L 1

Public Disclosure Authorized

Policy,Planning,and Research

WORKING PAPERS
Office
of theDirector

CountryEconomicsDepartment
The WorldBank
May1989
WPS231
Public Disclosure Authorized

The World Bank Revised

Minimum Standard Model
Public Disclosure Authorized

Conceptsand Issues

DougAddison
Public Disclosure Authorized

The Revised Minimum Standard Model was originally created

in 1973 to ensure a consistent approach to World Bank projec-
tions. Its primary purpose, like the original two-gap models, is
to show the user what levels of investment,imports, and external
borrowing willbe required to reach targetedreal GDP and export
growth rates. But the RMSMcannot provide guidance about the
policies or prices that would be needed to reach those levels.

The Policy, Planning, and Restarch Complex distributes PPR Woding Papers to disseminate the findings of work in progrcss and to
encourage the exchangeof ideas amongBank staif and aU others interested in dcvelopm,cntissues. Thescpapers carry the names of
the author., reflectonly their views, and should be used and ited accordingly.The findings, interpretations. and conclusions are the
authors'own. They should notbe attributedto theWorld Bank, its Boardof Directors,its management, or any of its manbercountries.
 Plc,Planning,and Research

Officeof the Director

The Revised Minimum Standard Model
(RMSM) was originally created in 1973 as a
means of ensuring a consistent approach to
World Bank projections and thus facilitate
intercountry comparisons. Thcse objectives are
met through the provision of a standard list of
variables and a minimum set of economic
relationships. The decision to minimize the
number of linkages recognizes two facts: the
quality of data for econometric analysis is
generally poor, and users undoubtedly want to
modify the model to meet their country-specific
needs.
The RMSM is a thinking and planning tool.
Its primary purpose, like the original two-gap
models, is to show the user what levels of in-
vestment, imports, and external borrowing will
be required for a targeted real GDP growth rate.
The planner's choice of a real growth rate will
determine what level of investment will be nec-
essary. If the RMSM is run as a trade-gap
model, the level of imports needed to sustain this
rate of growth is driven by GDP, investment,
and consumption - which is, in tum, residual.
If the RMSM is run as a two-gap model and the
savings constraint is binding, imports rather than
consumption are residual. Real export growth is
exogenous in both cases. The difference be-
tween exports and imports determines the need
for extemal borrowing. Thhisreflects the Bank's
"needs" or "requirements" approach, and
contrasts with the "constraints" or "availabili-
ties" approach, which determines the real rate of
GDP growth given available foreign capital.
The RMSM cannot, however, provide
guidance about the policies or prices needed to
reach the indicated levels. At first glance it
would seem that the model has a rather limited
scope. This is not so. The usefulness of any
model is determined by the questions asked of it.
One can easily and quickly discover whether or
not a targeted growth rate is acce- table in terms
of its impact upon per-capita consumption and
external financing needs. It is also easy to
manipulate the models' parameters to find out
how the economy might be restructured to make
a target growth rate practical. This sort of ex-
perimentation can lead to some very useful ob-
servations about the path a country should take
in the future.

This paper was prepared in the Africa Country Department IV with the advice and
support of the Country Economics Department staff. Copies are available free from
the World Bank, 1818 H Street NW, Washington DC 20433. Please contact Josie
Onwuemene-Kocha, room N1 1-032, extension 61750.

TdhePPR Working Paper Series disseminates the findings of woCkundeTway in the Bank's Policy, Planning, and Research
Complex. An objective of the series is to get these fuldings out quickly, even if presentations are less than fully polished.
'MTefindings, interpretations, and conclusions in these papers do not necessarily represent official policy of the Bank.

Produced at the PPR Dissemination Center

 The World Bank Revised Minimum Standard Model:
Concepts and Issues
by
Doug Addison

TABLE OF CONTENTS

I. INTRODUCTION ............ ....................................

1

II. 'TPHENATIONAL ACCOUNTS ....................................

2
1. Accounting Identities
2. Exogenous Variables
3. Investment
4. Import PFnctions
5. Income Effects and the Terms of Trade
6. The Constrained Savings Case
7. Dynamics
8. The Maximum Marginal Savings Rate

III. NTH BALACE OF PAYMENTS................................ 17

1. Accounting Identities
2. Exogenous Variables
3. The Resource Balance
4. External Borrowing
5. The Change in Reserves

IV. THE CHOICE OF PAUMETEmS ..................................

23
1. The Efficiency of Investment
2. Import Elasticities
3. The Marginal Savings Rate

V. SUMMARYAND CONCLUSIONS ................................ 25

APPENDIX A: DYNAMICS................................. 26

APPENDIX B: RMSM IDENTITIES 28

.............................

BIBLIOGRAPHY .................................... 29

*Research Analyst, West Africa Country Operations. Special thanks

to John Holsen, Bela Belassa and all those who contributed their
comments on earlier drafts.
 INTRODUCTION

1. The Revised Minimum StandardModel was originallycreated in

1973 as a means of ensuringa consistentapproach to World Bank projections
and thus facilitateinter-countrycomparisons. These objectivesare met
through the provision of a standardizedlist of variables and a minimum set of
economic relationships. The decision to minimize the number of linkageswas
made in recognitionof two facts: the quality of data availablefor
econometricanalysis is generallypoor and, secondly,users will undoubtedly
want to modify the model to meet their country-specificneeds. The number of
variables,unfortunately,is not at all limited: there are about 430
variables in the model today. Happily there is a subset of core variables
which form the backbone of the model. The remaindercan be left blank or can
be filled in according to the availabilityof information. It is the core
variables and equationsthat are describedhere in this paper.

2. The RMSM is a thinkingand glanning tool. Its primary purpose,

like the originaltwo-gap models, is to show the user what levels of
investment,imports,and externalborrowingwill be required for a targeted
real GDP growth rate.2 The planner's choice of a real growth rate will
determinewhat level of investmentwill be necessary. If the RMSM is run as a
trade-gapmodel, then the level of importsneeded to sustain this rate of
growth is driven by GDP, investment,and consumptionwhich is, in turn,
residual. If the TMSM is run as a two-gapmodel and the savings constraint is
binding, then importe rather than consumptionare residual. Real export
growth is exogenous in both cases. The differencebetween exports and imports
determinesthe need for externalborrowing. This is reflectiveof the Bank's
"needs"or "requirements"approach and stands in contrastto the "constraints"
or "availabilities"approachwhich determinesthe real rate of GDP growth
3
given available foreign capital.

3. On the other hand, the RMSM cannot provide any guidanceas to

the sorts of policies or prices that would be needed in order to reach the
indicatedlevels. At first glance it would seem that the model has a rather
limited scope. This is not so. The usefulnessof any model is determinedby
the questionsasked of it. One can easily and quickly discoverwhether or not
a targetedgrowth rate is acceptablein terms of its impact upon per-capita
consumptionand external financingneeds. It is also easy to manipulate the
models' parameters in order to find out how the economy might be restructured
in order to make a targeted growth rate practical. This sort of
experimentationcan lead to some very useful observationsabout the path a
country should take in the future.

See, for example, Chenery and Bruno (1962) or Chenery and Strout
(1966).

The original,discontinued,mainframeversion of the model was able

to accommodateeither method. The "availabilities"approach is not,
to this author'sknowledge,currentlyan option.
 2

THE NATIONAL AC.OUNTS

AccountingIdentities

4. The RMSM uses a two-gap accountingframework. It ensures

consistencybetween proiectionsof the Balance of Paymentsand the National
Accounts via the resource gap which both accountsshare in common. The
accountingidentitiesare: 4

GDP - C + I + X - M 1)
GDP - C + S 2)

I - S -M - X 3)

These identitiesare expressed in real terms where gross domesticproduct

(GDP) is the sum of %onsumption(C), investment(I) and exports (X) less
imports (M).5 Alternatively,GDP is the sum of consumptionplus saving (S).
Identity 3 states that tha two gaps between investmentand savings and between
imports and exports are the same. If the level of savings is inadequateto
finance a targetedlevel of investment,then investmentmust be financed from
abroad in the form of net imports. As shown in Section III below, this will
necessitateexternalborrowing.

ExogenousVariables

5. In the RMSM, real GDP growth is exogenousand must be targeted

by the user.6 There are two common stratsgies. The first is to choose a rate
which will ensure a constantor increasingGDP per capita growth rate in order
to find out what sort of investmentand externalborrowingwill be needed.
The second is to choose a level of growth that results in levels of investment
and borrowing that are consistentwith ones' expectationsof ability and
availability. This latter approach is a parallel to the "availabilities"
method and usually ends up as an iterativeprocess as growth rates are
adjusted to produce the expected feasiblelevels of borrowing and investment.

4. Please note that many variableshave been aggregatedfor the purpose

of this paper. The mnemonic rules for var'ablenames have,
however, been retained to make the transitionto or from the actual
model as painless as possible. Appendix B illustratesthe
translationin detail.

5. The national accountscan be expressedeither in local currency or

in US dollars. The assumptionused here is that local currencywill
be used.

6. GDP is disaggregatedinto three categories: agriculture,industry

and all other goods. Separategrowth rates are applied to each.
 3

6. Real export growth is also set by the user.7 This growth rate
describesexport sales rather than production. The user will have to reflect
all his or her assumptionsabout tradingpartner demand, market shares and
reactionsto changes in prices and excl'ange rates into this growth path.

GDPt - GDPt 1 * (1 + gdp_grt) 4)

Xt - Xt-l * (1 + x_gr ) 5)

There is no linkage of exports to GDP. Note that the growth of real export
sales can be temporarilyhigher than GDP growth only if there are substantial
stocks to draw down - or if the planner is willing to sacrificeconsumption
and/or investmentgrowth. 8

Investment

7. The RMSM investmentfunctionis a functionof desired GDP

growth. There are two methods of providing this linkage: a marginal
propensity to invest (MPI) or an incrementalcapital-to-outputratio (ICOR).
Equation 6, below, can accommodateboth.9

It kt + Alt * GDPt + A2t * (GDPt - GDPt-1) 6)

Parameter k representsautonomousinvestment. ParameterAl representsthe

marginal propensity to invest (MPI) when parameter A2 equals zero. Parameter
A2 representsan incrementalcapital-to-outputratio (ICOR) when parameterAl
equals zero.

8. This simple formulationleaves out many importantvariables.

Capacityutilizationand the applicationof labor are but two examples.
Although it is true that past performanceis not always an accuratepredictor
of the future, the modeler should endeavorto estimate these parameterson the
basis of historicalperformancebefore making any judgmentsabout what they
0
could become in the future.1

7. The user can specifyup to nine differentexport categories.

s. This can also include the recoveryof goods which pass through the
system of parallel markets.

9. The user can disaggregateGDP into three sectors: agriculture,

industry,and all other goods. The investmentfunction can also be
disaggregatedin the same way when the ICOR formulationis chosen.

. The author has made some rough estimatesof ICORs for several Sub-
Saharan African countries. The average was three plus or minus two
for high growth periods. The large variation is undoubtedly
reflectiveof the many economic forces left out of the investment
equation. (Unpublishedinternalnote, September 20, 1988, AF4CO)
 4

9. The reader may detect an anomaly in the subscriptsof the

investmentequation. Tne definitionof an incremental-capital-output-ratio
requires that the change in GDP for any given year be contrastedwith the
change in capital (investment)for the precedingyear. This is also
consistentwith the fact that the RMSM is a planningmodel: future GDP growth
should determirnepresent investment. Equation 6, therefore,should actually
read:

It - kt + Alt * GDPt + A2t * (GDP..1 - GDPO) 7)

The fact that it does not was a deliberatemis-statementon the part of the
designers of the current RMSM. This was done because they did not want the
last year of the projectionsto be incompletebecause of a lack of
information. The original RMSM manual states that the differencein results
is minimal. This is true with regard to general trends and magnitudes. It
could be misleading,though, if one is concernedwith the timing of investment
relative to targetedgrowth. A simple solutionwould be to specify the growth
target for one year beyond the last year for which investmentis to be
projected.

Import Functions

10. The demand for imports is assumed to be driven by the need for
consumption,intermediateand investmentgoods. The demand functionsare
expressed using growth rates computedas the rate of clange in each of the
aggregatesmultipliedby import elasticitiesof demand. Note that these are
not price or income elasticitiesof demand. They are nothing more than simple
linkages to the aggregate supply and use of goods. They are sometimes
referred to as "composite"elasticitiesbecause their value must reflect the
combined effects of prices, income, quantitativerestrictionsand exchange
rate constraintsas well as other demand related factors. None of these
variables enter explicitlyinto the model.

11. There are three separateequations: imports of intermediate

goods (Mg) are tied to changes in GDP via elasticitymg_el, importsof
investmentgoods (Mi) are linked to changes in investmentusing elasticity
mi_el, and consumer imports (Mc) are coupled to consumptionby means of
1
elasticitymc_el."

Mgt = Mgt-,* (1 + mg elt * ( - 1)} 8)

GDPt.
1

Mit - Mit.1 * (1 + mi elt * ( I - 1)) 9)

. The RMSM actually includesseveral import categories. Intermediate

goods, fuels, and non-factorservicesare assumed tied to GDP.
Capital goods are tied to investment. Food and other importsare
linked to personal consumption.
 5

mct - Mtctl * (1 + mc elt * Ct 10)

Ct-1

12. Although the growth rate analogy is a very intuitiveway to

think of import demand, one can gain greater insight by viewing the change in
import demand as being regulatedb-rshare-weighted-import-elasticities. The
phrase "share-weighted-import-elasticities" includesboth the share of imports
supplying a particulareconomicactivityand the elasticityof demand for that
type of imports. For example,how .ch investmentis sustainedby imported
capital and how does the need change as investmentchanges? The import
functions are rewrittenbelow using this method.

Mgt - Mgt-,* (1 - mg_el,) + mg_elt * Mgt-, * GDPt 8)

Iit - Mit-1* (1 - mi ,el_) + mi elt * Mit *I 9)

it-1

mct - Mct- 1 * (l-mc_el,) + mcel, * MC.-I* ct 10)

Ct-1

The advantageof the weighted elasticitiesapproach is that it allows one to

learn somethingabout the directionof change and the share of imports
relativeto the other aggregates. The shares approach is also useful because
it encouragesone to think of the need to import as an evolving strtctural
feature. For example,one cou'd create a situationwhere a growing economy
obtains more of its own capital domesticallyby making the investment-imports
elasticityless than one. The followingset of rules, embodied in Table 1,
will prove to be a useful guide. It is assumed that GDP growth is positive.

Table 1

Elasticity Real Import Growth Imnort Shares

m_el > 1 positive increasing

m_el = 1 positive constant

0 < m_el < 1 positive declining

m el < 0 negative declining

13. At this point all but one variablehas been described. That
variable is consumption. The closure rule is:

Ct - GDPt - It - Xt + Mt 11)
 6

We will pay special attention to the effects of CDP and export growth, as well
as tOe choice of parameters,on consumption. Real growth in consumption,per-
capita consumptionin particular,is perhaps the most importantcriteria in
assessing the acceptabilityof the growth scenariobeing modeled. Growth
which does not increase real per-capitaconsumptionis rarely politically
sustainable.

Income Effects and the Terms of Trade

14. Up to this point, the discussionhas omitted the effects of the

terms of trade. The present formulationof the accountingframeworkcorrectly
implies that one must supplementdomestic output with net imports (M-X)
insofar as expendituresexceed total domesticoutput (GDP). It says nothing
about the gains or losses resultingfrom changes in the terms of trade. In
the social accountingframeworkused with the RMSM, however, we take these
gains or losses into account by making a distinctionbetween gross domestic
income (GDY) and gross domesticproduct (GDP) -- the differencebetween the
two being the income effects of changes in the terms of trade:

GDY, - GDP, + TTADJt 12)

where TTADJ stands for the adjustmentto GDP for changes in the terms of
trade.

15. In calculatingrea'lGDY, we value productionfor exports in accord

with the resulting capacityto import. The capacity to import from exports is
calculatedby deflating .3xportearnings (exportsat current prices) by the
price of imports. In this way, we take into account price changes measured
against the base year for both exports and imports. The capacity to import,
in other words, is equal to exports adjusted for the terms of trade (XTTADJ).
It is the equivalentof lookingupon exportingas an indirectway of
"producing"imports.

XTTADJt - (Xt * XPId)/MPIt 13)

Note that XPI and MPI are price indices expressedin local currency terms.12
These indices should share the same base year as the rest of the national
accountsvariables.

12. The terms of trade adjustment is independent of exchange rate

movements because the formula results are always denominated in the
same currency as the national accounts.
 7

16. The income effect of changes in the terms of trade (TTADJ) thus is
the differencebetween the capacityto import and real exports as ordinarily
calculated. It is written as:

TTADJt - XTTADJt - Xt

- (xt * XPIt)/MPIt - X; 14)

The terms of trade adjustmentis simply a way of taking into account the
effects of changes in export and import pr'ces upon potentialreal
expenditures. If the export and import price indices are equal, the
adjustmentwill be zero. In the base year, therefore,the terms of trade
adjustmentwill always equal zero and GDP will equal GDY. If the terms of
trade have improved (XPI > XPI) since the base year, then the adjustmentwill
be positive (and vice versa).

17. Domestic saving must also be adjusted for the income effects of
changes in the terms of trade. Domestic saving is properlydefined as income
less consumptionrather than as productionless consumption.

GDSt - GDYt - Ct 15)

This is identicalto investmentless the resourcegap (RG) where the resource

gap is defined as exports adjustedfor the terms of trade less imports.

GDSt - i RGt 15)

RGt - Mt- XTTADJt 16)

Note that consumptionand investmentare not affectedby the introductionof

the terms of trade adjustment. Equation11, however, can be re-writtenas:

ct - GDYt + RGt - It 11)

The ConstrainedSavings Case

18. The RMSM is often run as a one-gap (trade gap) model. When
this is done, the model is closed as indicatedin paragraph 13 and equation
11: consumption is a residual. Savings is also a residualsince it is
defined as income less consumption. The RMSM, however, was originally
intendedto be, and can scill be used as, a two-gap model in which either the
trade or the savings gap may prove to be the binding constraint. Particularly
where there are socio-politicalconstraintson consumptionand spending, it is
desirable to specify a realisticmaximum marginal savings rate. This is the
same as choosinga minimum marginalpropensityto consume. When the savings
rate binds, net importsmust exceed the level requiredto fill the trade gap
because it is necessary to have sufficientresourcesavailable to meet both
required investmentand minimum consumptio'x levels.

19. To see if the savings gap (ratherthan the trade gap) is

binding, one must solve for the level of importsnecessary to guarantee the
 8
13
maximum marginal savings rate (HAXMSR)is not exceeded. This is done below:

Target M, - It + XTTADJt - GDSt_. - MAXMSRt * (GDPt - GDPt-1) 17)

In order to reach this targetedlevel, one could calculatethe difference

between total imports (Mg + Mi + Mc) and the targetedlevel of imports. This
differencecould then be added to the total. In practice,however, it is
assumed that the adjustingvariablewill be consumption-related imports. The
motivation for making this adjustmentwas, after all, the desire for
politically acceptablemarginalsavings and consumptionrates. If total
imports, including (Mc), produce an overly high savings rate then (Mc) will be
re-calculatedto produce a MSR at the maximum acceptablelevel.

Mct - It + XTTADJt - Mgt - Mit - GDS,1 - MAXMSRt * (GDPt-GDPt-

1) 18)

when a MAXMSR is specifiedand Mc. > Mct as defined in Equation 10

20. Keep in mind, when using the MAXMSR, that using imports to
maintain consumption in the face of increasedinvestmentneeds may not be
sensible if it results in an overly high level of foreignborrowing.
Financingconstraintsmay render the maximum savings option moot. Keep in
mind also that boosting importsto compensateincreasedexports is counter-
productivewhen a nation needs to generatea trade surplus in order to pay off
its foreign obligations. 'he system of equationsin real terms is now
complete and summarizedin Tabie 2 below.

13. The actual RMSM formula ties importsof consumptiongoods to gross

national rather than domesticincome and savings. The difference
between the two conceptsis factor service income and private
unrequited transfers.
 9

Table 2

The Revised Minimum StandardModel: Condensed Form

Exogenous:

GDP: GDP,.- GDPt-.* (1 + gdp_grt) 4)

Exports: Xt - Xt_ * (1 + x gr-) 5)

Endogenous:

Investment: It - kt + (Alt+ A2t) * GDPt - A2t * GDP,- 6)

Imports

IIntermediate:
Mg, - Mgt-,* (1 + mg_elt * (----; ;- - 1)) 8)

1 * (1 + mi elt * (.I.. ;
. Investment: Mit - Mit. 1)} 9)
itI
Consumption: Mct - Mct1 * (1 + mc elt
C * (--$ - 1)) 10)
Ct-1

or... Mct - It+XTTADJt-Mgt-Mit-GDSt-1 -MAXMSRt*(GDPt-GDPt

,) 18)
when a MAXMSR is specified and Mct > Mct as defined in Eq. 10

Terms of Trade Revaluations:

Terms of Trade: TTADJt - (Xt * XPIt)/MPIt - Xt 14)

GDY: GDYt - GDPt + TTADJt 12)

Exports: XTTADJt Xt + TTADJt 13)

Resource Gap: RGt - - XTTADJt 16)

Savings: GDSt - It - RGt 15)

Closure

Consumption: Ct - GDPt - It - Xt + Mg, + Mit + Mct 11)

- GDYt + RGt - It
 10

Chart 1

'1
x_gr-> X

gdp_gr-> GDP Il

m~~

*1**

L-LIJI4I~
 11

Dynamics
21. In this section of the paper we will investigatethe effects of
changes in real GDP and export growth on each of the variables. For
illustrativepurposes, it will be assumed that we are working with an economy
that has very inefficientcapital and high import elasticitiesof demand. If
you are mathematicallyinclined,you may want to refer to Appendix A which
containsthe equationsupon which this discussionis based. Please note that
the results shown in the followingfigures are extreme cases chosen for
expositorypurposes: the modeler would not, under ordinary circumstances,
expect to see such exaggeratedresults.

22. We will start the analysiswith investment. It was explained

earlier that the modeler has two ways of linking investmentto GDP: an
marginal propensityto invest (Al) or an ICOR (A2). Examplesof both can be
seen in Figures 1 and 2 below. If the MPI (Al) is used, then the trend in
investmentmerely mimics the trend in GDP. If an ICOR (A2) is used, then
investmentincreasesboth with the absolute level of GDP and with the rate of
growth in GDP. That is why the effect is so much more volatile in Figure 2.
(See AppendixA.1)

23. Importswill move in the same directionas, and parallel to the

trends of, the variables they are linked to so long as the elasticitiesof
demand are positive. The share of these imports in economicactivity,
however, will change unless the elasticitiesare set equal to one. (See Table
One.) Figures 3 and 4 are examplesof intermediateand capital goods imports
during a short burst of GDP growth when the respectiveelasticitiesof demand
are larger than one and investmentis linked to GDP via an ICOR. These will
be the standardassumptionsin the examples that follow. (See Appendix A.2
and A.3)

24. Consumptionand consumption-related imports (Mc) are

interdependent. They are also the only aggregateswhich are linked to the
level of exports as well as GDP. Considerconsumptionfirst. Positive
changes in GDP or importswill add to consumption;increases in investmentor
exports will take away from it. The total effect of GDP growth is therefore
mixed. The final result will be determinedby the efficiencyof investment,
the strengthof the import elasticitiesand the starting import shares. The
overall effect,however, will almost always produce a marginal propensityto
consume greater than zero and less than one - thus producingthe traditional
positive relationshipbetween income and consumption. (See AppendixA.4.)
 12

C)~~~~~~~~~~~~~~~~~~~O
1.4 u

1-4
r4 GD

Time
Time I I I...
I I I.7. ,

Figure 1: Investmentand GDP Figure 2: Investmentand GDP

I -Al * GDP I - A2 * A GDP

w ~~~~~GDP?1
f**4 '~~~~~~~~~~~~~~~~~~~~~~~~~~-4I

Mi

Time ~~~~~~~~~~Time

Figure 3: Intermediate Figure 4: Capital Imports

Imports I - A2* A GD?
 13

25. Consumptioncan be made to fall, even though GDP is growing,

under two conditions. The first, and most obvious,of these would be a large
increasein exports. The second case would occur when consumptionis
sacrificedfor investment. This could be inducedby an accelerationin GDP
growth combinedwith a large inefficientICOR and an elasticityof less than
one for imports of investmentgoods. The effect of the low import elasticity
is that imports will fail to increaserapidly enough to compensatefor the
large expansion in investmentspendingneeded to support the increasedGDP
growth: consumptionmust thereforefall. Figure 5 illustratesthis case.
Imports of consumptiongoods, as shown in Figure 6, will mirror the trend in
consumption. (AppendixA.5)

~~~~~~Tm TimeX

Figure 5: Consumptionand GDP Figure 6: ConsumptionImports

 14

26. Figure 7 shows what would happen to the resourcegap, without

any terms of trade adjustments,in the inefficientICOR scenarioused in
figures 2-4 above. An increasein GDP growth will require more capital goods
which, in turn, will necessitatemore capital goods imports since, in this
example, the capital importselasticityis greater than one. The resource gap
can be reduced by increasingthe growth of exports,by using lower import
elasticities,or by increasingthe price of exports relative to the price of
imports. (See AppendixA.6) Whether any of these changes would in fact be
reasonablein a particularcase is, of course, a matter of judgement.

U GD

r;=

Figure 7: The Resource Gap

 15

27. Up until this point we have examinedwhat happens when the

savings and consumptionfunctionsare unconstrained. If the user has decided
to use the maximum marginal savings rate (MAXMSR),then the above equations
will hold until the marginalmaximum savings rate is reached. At that time,
the followingequationswill take effect:

Ct -Ct-1 + (1 - MSR) * (GDPt - GDPt1) 19)

Mt-It + XTTADJt -GDS,_ 1 - MAXMSRt* (GDPt - GDPt 1) 17)

RGt -It - GDSt_. - MAXMSRt* (GDPt-GDP, 1 ) 20)

AGDS, MAXMSRt 21)

AGDPt

AC- - 1 - MAXMSR, 22)

4!L., / AL. + /AXTTADJI. MAXI4SRt 23)

AGDPt AGDPt A\GDPt

ARG, AI, MAXMSRt 24)

,/\GDPt LACDPt AMR

GDP growth, under this constraint, will require more external borrowing as
long as the marginal savings rate is less than the marginal propensity to
invest. Observe that export performancedoes not enter into the resource gap
equation at all. The reason, as expressed in equation 17, is that exports are
prevented from competingwith consumptionthrough the provisionof more
imports. Put another way, the marginal savings rate is a binding constraint
on the savings and consumptionfunctions. Changes in investmentor exports
will, therefore,be compensatedby the level of imports. Figure 8 below shows
the effects of increasedGDP growth with the high ICOR and import elasticity
scenarioused above. Figure 9 illustratesthe effects of boosting export
growth.

28. Note that it may be unrealisticto produce a situationwhere an

increase in exports automaticallyincreasesimports: such a result would be
viewed as counter-productive when a trade surplus is needed in order to pay
off foreign obligations. It would be a better policy to maintain consumption
by reducing inefficientinvestmentflows and replacing them with more
productiveadditions to the stock of capital.
 16

U / 42 x

'-4~~~~~~~~~~~~~~~~~~~~~~~~~~4

C~~~~~~~~~~~~~C

Time Time
114 1~1 I I I I II I IIIII . 11 l1111111 II

Figure ConstrainedSavings Figure 9: ConstrainedSavings

with GDP Growth with Export Growth
 17

THE BALANCE OF PAYMENTS

AccountingIdentities

29. The standardpractice in balance of payments presentationsis

to show the current account (CURBAL)balancedby the capital account (NETCAP)
plus the change in reserves (CHGRES). Each of the accountsare expressed in
net nominal US dollars. If, for example, there are more dollars flowing out
of the capital account than in, then NETCAP will be negative and the sum of
the current and reserve accountswill positive.

CURBAL + NETCAP + CHGRES - 0 25)

The current account includesthe resourcebalance (RESBAL),net factor

services (NETFSY)and net current transfers (NETCTR). The word "current,"
with respect to transfers,is meant to distinguishthese transfers from
official transferswhich are recordedin the capital account and used for
investmentpurposes.

CURBAL - RESBAL + NETFSY + NETCTR 26)

The resourcebalance (RESBAL)is defined as exports (EXP) less imports (IMP)

of goods and non-factorservices. As will be shown below, it is the resource
balance that provides the linkagebetween the national accountsand the
balance of payments.

RESBAL - EXP - IMP 27)

30. A short digressionis now necessary. Both the current and the
capital account contain distinctionsbetween foreign exchange flows set into
motion by events prior to the projectionperiod and those flows generated
during the projectionperiod. The former are referredto here as "pipeline"
flows. The latter flows can be grouped into two types: exnectednew flows
and additionalunidentifiedflows needed to balance the accounts. These last
are referred to as GAPFIL flows. These distinctionsare importantbecause
they place emphasis on the questionof how much more money a country will have
to borrow given existingpipelineand expectednew flows.

31. Factor service income (NETFSY)will be defined here as interest

accrued on GAPFIL borrowing (INTGAP)plus all other factor services (OTHFSY).
This last item includespipelineand expectedinterestflows, net payments on
direct foreign investmentand net income from property and labor.

NETFSY - INTGAP + OTHFSY 28)

32. The capital account contains net GAPFIL flows (NETGAP), net
expected and pipeline long-term flows (LTCAPF) and other capital flows
(OTHCAP). Net flows here are defined as disbursementsand other credits less
repayments and other debits.
 18

NETCAP - NETGAP + LTCAPF + OTHCAP 29)

NETGAP - GAPFIL - AMTGAP 30)

Net long-term capital flows (LTCAPF)include net direct foreign investment

(NETDFI),official capital transfersor grants (DBTGRT),net long-term
borrowing (NETLT) and other long-termflows (OTHLTF). Other capital flows
(OTHCAP) include net short-termcapital (SHTERM),capital not elsewhere
included (CAPNEI)plus errors and omissions (ERROMS).

ExogenousVariables

33. The following items are determinedexogenously: other factor

services (OTHFSY),net current transfers (NETCTR),net long-termflows
(LTCAPF)and net other capital flows (OTHCAP). Note that net long-term
borrowing (NETLT) is made up of both expectedand pipeline flows. Pipeline
flows are usually calculatedby the World Bank's Debtor Reporting System from
informationabout existing loans. Expected flows are computed exogenouslyin
a separate debt module based on the modelers assumptionsabout new loan
commitmentsand terms. All other factor service and net capital items as well
as net current transfers are projectedby means of exogenousgrowth rates.
These growth rates are expressed in nominal terms and so must includeboth
value and volume effects.

The Resource Balance

34. Exports (EXP) are equal to real exports in local currency (X)
multipliedby an exogenous export price index (XPI) and exchangerate (E).
The same relationshipholds true for imports.

RESBAL G EXP - IMP 27)

EXP - X * XPI * E 31)

IMP M * M1P * E 32)

Although the exchangerate and price indices affect the valuation of the
resourcebalance, they do not induce any changes in real trade volumes. The
overall trends are all determinedon the real side of the model. It is there
that the modeler must imDute reactionsto prices and exchangerates by
manipulatingthe real export growth rate and the import elasticities.

14. Many modelers choose to use the World Bank's Half Yearly Commodity
Price Forecast for the export and import price indices. These are
based on global supply and demand projectionswhich include
assumptionsabout large producersand consumersfor certain
commoditiesand countries. The modeler may treat each country as a
"price taker" when these indices are used.
 19

The Change in Reserves

35. The assumptionmade by the RMSM authorswas that policy makers

should try to maintain a level of reservesadequate to finance two months of
imports. Each incrementalchange to the level of importswill therefore
require an additionalone sixth that amount to be set aside as reserves --
assuming the stock of reservesis already adequateat the beginning of the
15
projectionperiod.

CHGRES - (IMPt - IMPt 1) * (1,) 33)

If the starting level of reserves is insufficient, the modeler may want to

modify the formula to calculatehow much more should be added to the stock of
reservesover the first severalyears of the projectionperiod. Keep in mind
also that governmentswith fixed exchange rate systems are likely to be more
concernedwith the level of reservesthan those with market systems.

ExternalBorrowing

36. The need for *additional

unidentifiedborrowing (GAPFIL)is a
residual. It is determinedin part by the size of the resourcebalance. This
pressure is further amplifiedby the change in reserves (CHGRES),which is
directly linked to imports,and by the need to repay previousdebts (AMTGAP)
?lus interest (INTGAP)on the stock of outstandingdebt (DODGAP). Exogenous
forces are representedby other factor service income (OTHRSY),net current
transfers (NETCTR)and net pipelineborrowing (NET).

GAPFIL - - RESBAL - INTGAP - OTHFSY - NETCTR +

- AMTGAP - LTCAPF - OTHCAP - CHGRES

- - CURBAL - AMTGAP - LTCAPF - OTHCAP - CHGRES 34)

It is assumed that borrowing in one period will generate a repayment stream in

the future: there is no provision for building up arrears as so often happens
6
in real life.1 For the purpose of this discussionit is assumed that all

15. The RKSM also assumes that there will be interestearned on foreign
exchange reserveholdings abroad. These earnings are includedas
part of OTHFSY. They were not made explicitbecause their inclusion
does not significantlychange the dynamicsof the model.
 20
loans are repaid in the period followingdisbursement.

AMTGAPt- GAPFILtl 35)

Next, assume that interest (INTGAP)is charged and paid on the stock of
outstandingdebt (DODGAP). Note that GAPFIL borrowing in each period
increases interestcharges in the next period which further increasesthe need
for GAPFIL financing.

DODGAPt- DODGAPt_
1 + GAPFILt - AMTGAPt 36)

INTGAPt - DODGAPt-
1 * it 37)

37. GAPFIL borrowing is normally assumed to be made on commercial

bank terms for non-IDA countries. Because of this, the modeler should seek to
ensure that the level of this costly borrowing is not unreasonable. A useful
gauge of what is and is not acceptablefor total borrowing is the debt service
ratio. A debt service to exports ratio of less than 20 or 25 percent is
generally consideredreasonablealthoughthis may be temporarilyexceededfrom
time to time. If there is too much GAPFIL borrowing the RMSM user should
review his or her assumptionsabout the terms and amounts of the other flows in
the model. Perhaps more borrowing could be shifted into concessionalsources.
If there is not sufficientflexibilitythere, then it is possible that the
targets for real GDP and export growth will have to be revised.

38. Note also that the RMSM will generatea negativeGAPFIL when
there is a surplus of funding. 'Thiswill also produce negative interest
(meaningthe interestis earned rather than owed) which will feed back into the
GAPFIL equation in the next period. Assuming that no mistakes were made with
regard to other variables, the user must then make some decisions about how the
earnings are to be allocated. One use would be to increasereserves - although
this will require a modificationto the formula for CHGRES. One could also
eliminate the excess by importingmore or borrowing less.
 21

39. All BOP variableshave now been defined. The complete system of
equationsis shown in Table 3. The behaviorof the balance of payments system,
as a whole, is uncomplicated. Most of what happens is already predeterminedby
the rnlationshipsin the nationalaccounts. The only tricky part is the feed-
back loop between interestcharges and the size of the CAPFIL.17

17 There is an additionalfeed-backloop betfeen the balance of

paymentsand the nationalaccountswhich .- ught to be discussed. An
earlier footnoteindicatedthat importsof consumptiongoods, when
under the maximum marginal savings constra.tt,are actually linked
to gross national savings rather than gros.edomestic savings. The
latter do not include factor service income or transfers. This is
an importantomissionbecause factor service income includes
interestpaymentswhich expand or contractwith the amount borrowed.
If the MAXMSR is used, then the effects of a real resourcegap (M >
X) will be exacerbatedby the interestcharged on the resulting
externalborrowing.
 22

Table 3

The RMSM Balance of Payments: Condensed Form (Nominal USS)

Exogenous:

Export Price Index XPIt

Import Price Index MPIt

Exchange Rate: Et

Other Factor Services: OTHFSY,

Net Private Transfers: NETCTRt

Long-term and Other Capital: LTCAPFt and OTHC'4Pt

Endogenous:

Resource Balance: , RESBALt - EXPt - IMPt 27)

. Exports, Goods and NFS: EXPt - Xt * XPIt * Et 31)

IImports, Goods and NFS: IMPt -M * MPIt * Et 32)

Net Factor Services: NETFSYt - INTGAPt + OTHFSYt 28)

IInterest, New Borrowing: INTGAPt - DODGAPt1 * it 37)

. Debt Outstanding (stock) DODGAPt - DODGAPt, + GAPFILt - AHTGAPt 36)

Current Account Balance: CURBAL, - RESBALt + NETFSYt + NETCTR, 26)

Capital Account Balance: NETCAPt - NETGAPt + LTCAPFt + OTHCAPt 29)

. Amortization, New Borrowing: AMTGAPt - GAPFILt- 35)

Change in Reserves: CHGRES, - (IMPt - IMPt1) * (1/6) 33)

Closure:

Unidentified GAPFILt - - CURBALt - AKTGAPt + ...

New Borrowing: - LTCAPF - OTHCAPt - CHGRESt 34)
 23

THE CHOICE OF PARAMETERS

40. Having settled on a set of real growth rates for GDP and
exports, the planner may wish to manipulatethe RMSM parametersin order to
obtain more realistic or acceptableresults. It could be, for example, that a
targeted increase in real growth rate of per capita consumptionwould result in
an unsustainablelevel of externalborrowing. If this were the case, then the
planner might want to simulatethe effects of a structuraladjustment.
Structuraladjustment,in this case, implies changes in the efficiencyof
investment,the efficiencyof import usage, the marginal savings rate or
changes in the share of each national accountsaggregate in the general supply
and use of goods.

The Efficiencyof Investment

41. One way to ensure that adequateconsumptionis maintainedand to

decrease the need for externalborrowing is to improve the efficiencyof
irnvestmentso that less capital can make a greater contributionto output.
Less investmentmeans, ceteris paribus, that more domesticoutput and imports
can go towards consumption. If the share of imports in investmentremains -

constant then the demand for capital importswill decline and so will the need
for external finance. This could be done by increasingthe capacity
utilizationof under-usedfactories,the applicationof a larger labor force to
existing capital,better managementof existing capital,or the purchase of
more efficientcapital. The modeler can reflect these improvementsby using
smaller investmentparameters. This will be true no matter which functional
form the modeler has chosen for the investmentfunction. Note that these are
all essentiallyplanning options. The necessary incentivesto make them
successfulcannot be modelled in the RMSM.

Import Elasticities

42. Another way to reduce borrowing requirementsis to directly

reduce the dependencyupon imports. Note that a reduction in imports could
also have a harmful effect by cutting into consumptionif investmentor exports
are not reduced in compensation. The criticalquestion is whether or not the
average long-runelasticitiesare to produce expandingor shrinking import
shares. For example, if the modeler wants to assume an economy which becomes
progressivelybetter at satisfyingconsumeror investmentneeds out of domestic
8 (See Table 1
production,then the elasticitiesshould be less than one.i
above.)

18. An empirical study by Lopez and Thomas finds that the long-term
composite elasticityfor total imports should range from about 1.1
to 1.4 with higher coefficientsduring high growth periods. They
found very few countrieswith elasticitiesof less than one for
sustainedperiods.
 24

43. The RMSM user can also allocate the change in parametersover
the three types of import functions: intermediategoods, capital goods and
consumptiongoods. The relativeeffect of each is determinedby the starting
value of the respectiveimport shares. Keep in mind, also, that the capital
goods import function can be especiallyvolatile if the investmentfunction
uses an ICOR rather than a MPI or a lagged function.

The MarginalSavings Rate

44. If the modeler has imposed the MAXMSR constrainton savings then
he or she has ensured a protected level of consumptionat the expense of the
need to borrow. Once the maximum marginalsavings rate has been reached the
need for borrowing can only be reduced if the marginal propensityto invest can
be brought below the MAXMSR as illustratedby equation24.
 25

SUMMARY AND CONCLUSIONS

45. The RMSM is primarilya planning tool and thinking aid. It

calculatesthe amount of investment,imports,and externalborrowing that will
be needed in order to sustain a targetedreal growth rate in GDP and exports.
Except in the constrainedsavings case, consumptionand savings are residual
items which float up or down as a consequenceof the relative changes in the
other nationalaccounts aggregates. If consumptiondeclinesor if external
borrowing increases too quickly,this is a sign that the desired rate of growth
is unacceptable- or that structuralchanges are needed before society can
sustain that amount of growth without damage.

46. Structuralchange, defined as a shift in the compositionof the

supply and use of goods, is extremelyeasy to manipulatewithin the model. The
amount of investmentneeded to fuel domesticoutput is controlledthrough a
marginalpropensityto invest or an ICOR. The level of imports needed to
support investment,consumptionor intermediateuses is controlledthrough a
set of demand elasticities. Export performanceis controlleddirectlyby means
of a real growth rate. If desired,marginalchanges in consumptionand savings
rates can be constrainedwithin politicallyacceptableboundaries.

47. The clarity of the model is its main strengthbut also a source
of weakness. The simplicityof the linkages in the RMSM, the lack of prices
and other key policy variables,and the number of exogenousvariables presents
the modeler with numerous challenges. The planner can describe the desired
changes and trends but he or she cannot be sure that these will be consistent
with the existing arrangementof price and other incentivesin the economy. In
consequence,the planner cannot indicatehow prices and policies should be
changed.19

l9. This charge is not unique to the RMSM but holds for all similarly
constructedtwo-gap models, includingthe original, as well as many
other types of planningmodels. See for example the comments of
Kindlebergerand Herrick (1977).
 26

APPENDIX A

This appendix containsa list of the real marginal changes in

each of the endogenousvariables of the national accountsper marginal change
in GDP and exports.

A.1. Investment

Ait- - Akt g+
4'Al" + Alt + A2t
~ - A21 * 1 1
(1l+gdp_grt..
AGDPt AGDP, gdp_grt gdp_gi. gdp_grt* 1)

- Alt + A2t * gt

when parametersAl and A2 are constants,where:

=1 1 + 1
gdpgrt gr dp_grt * gdp_grt-.

For the sorts of growth rates normallyexperienced,the lagged GDP growth

function (g) is only slightlysmaller than GDP grow.h. Thus, one can think of
the two as being essentiallythe same.

If the parametersare constants then the average propensityto invest will

equal the marginal propensityto invest. If A2 equals zero then Al will equal
the average and marginal propensityto invest. If Al equals zero then the
real rate of GDP growth multipliedby the ICOR (A2) will equal the average and
marginal propensitiesto invest.

A.2. Imports. IntermediateGoods

4\Mgt = mg elt * Mgp3

/\GDPt GDPt..
1

= sgt

where sgt representsthe share-weightedelasticityfor imports of intermediate

goods.

A.3. Imports.Capital Goods

A\Mit mielt* Mit * It

\GDPt Iit-,- GDPt

= si si* AGDPt
*
\It

where si represents the share-weighted elasticity for imports of capital

goods.

Continued ...
 27

A.4. Consumption

The marginal propensityto consume,within the RMSM, is expressedas:

/\Ct1 * l((sgt + 1) + (sit - 1) * AG, - AG )

,aGDPt I -sct GD 4GP

-L - -1
-
AX,
The parameters sg and si are describedabove. Parameter sc is short-handfor
the share-weightedelasticityfor importsof consumptiongoods.

sct - mc elt * Ct1

Note that it is possible, though highly improbable,that an economy could

import nearly all of its consumptiongoods and do so with an elasticitywith
an absolutevalue greater than one. If this were the case, then the left-hand
side of the product in the marginal propensityto consume would be negative.
The overall marginal propensityto consume will, however, almost always be
positive and less than one.

A.5. Imports. ConsumptionGoods

Imports of consumptiongoods, as illustratedbelow, will simply mirror the

trend in consumption.

AMct sC * AC,,
AGDPt -I-set \GDPt

AM, - -set
Axt

A.6. The Resource Gap

The resourcegap will generallyincreasewith GDP growth through the effects

of the increasedneed for imports. If, however, elasticitysc outweighs
elasticitysi and if the product of the differenceand the marginalpropensity
to invest outweigh elasticitysg, then GDP growth could induce a reductionof
the resource gap. This is most easily achievedwhen all three elasticities
are negative. An increasein exports will also reduce the resource gap.

ARG - 1 sXTTADJt Ai, -

lat* (sg.,
+ sct + (Si, - sct) * AGP.AGP
 28

APPENDIX B

Anyone who has already used the RMSM will want to know how the
condensed model variables are related to those in the actual model. The
condensed version of the RMSM can be translated to the actual model by using
the rules below. For a complete description of these terms and of how they
are calculated, see the User's Guide to the Revised Minimum Standard Model.

B.l1.The National Accounts

GDP - YAGR + YIND + YOTH + INDTAX

X - XGNFS

I - IFIXED + CHGSTK

Mg - MINT + MEET + MNFS

Mi - MCAP

Mc - MFOOD + MOCG

RG - MGNFS - XTTADJ

where MGNFS - Mg + Mi + Mc

B.2. The Balance of Payments

RESBAL - EXP - IMP

EXP - EXPCM1 through EXPCM6 + EXPMAN + XPNFS

IMP - IMPFOOD + IMPOCG + IMPPET + IMPCAP + IMPINT + IMPNFS

NETFSY - INTGAP + OTHFSY

OTHFSY - DIIP + INTRES + INTLT + INTSTP + IMFCHG + OTHFSR - OTHFSP

NETCTR -WRKRMR - WRKRMP + OTHCTR - OTHCTP

CURBAL - RESBAL + NETFSY + NETCTR

NETCAP - GAPF1L + LTCAPF + OTHCAP

CHGRES - NETIMF + OTHCHR

 29

BIBLIOGRAPHY

1. Chenery, H.B. and M. Bruno, Development Alternatives in an Open Economy:

The Case of Israel, The Economic Journal, March 1962.

2. and A. Strout: Foreign Assistance and Economic Development,

American Economic Review, 1966.

3. Note 4.00, Statistical Manual, The World Bank

4. Holsen, J.A., The "Minimum Standard Model" for CPP System and Global
Framework Country Economic Projections. Country Economic Projections
Department, The World Bank, January 1973.

5. Balance of Payments Manual, International Monetary Fund, Second Edition,

1977.

6. Kindleberger, C.P. and B. Herrick, Economic Development, McGraw-Hill, New

York, 1977, p. 364.

7. The Revised Minimum Standard Model, Comparative Analysis and Projections

Division, Economic Analysis and Projections Department, July 1979
(Revised June 1988).

8. User's Guide to the Revised Minimum Standard Model, Country Economics

Department, June 1988.

9. Import-Growth Relationship and Policy: Some Considerations for Africa,

Lope; and Thomas, Trade Policy Division, Country Economics Department,
March 7, 1988.
 PPR Working Paper Series

Title Author Date Contact

WPS208 Effec*ive Primary Level Science

Teaching In the Philippines Marlaine E. Lockheed May 1989 C. Cristobal
Josefina Fonacier 33640
LeonardJ. Bianchi

WPS209 Can the Industelel Countries

Return to Rapid Growth? Space
Seminar Internatlonal
Economics Department
and International
Economic Analysis and
Prospects Division

WPS210 Notes on Cash-Flow Taxation Roger H. Gordon

WPS211 Price Policies for the Coffee

Sector In the Dominican Republic Panos Varangis

WPS212 Beyond the Debt Crisis: Aiternative

Forms of Financing Lessard

WPS213 Conditionality and Debt Relief Stijn Claossens

Ishac Diwan

WPS214 Macroeconomic Adjustment and Labor

Market Response: A Revlew of the
Recent Experience In LOCs Peter R. Fallon
Luis A. Riveros

WPS21S Adjustment and Income Distribution:

A Counterfactual Analysis Francois Bourguignon
WilliamH. Branson
Jaime de Molo

WPS216 Price and Quality Effects of

Vers Revisited: A Case Study of
Korean Footwear Exports Jaime de Molo
L. Alan Winters

WPS217 Public Debt, North and South Helmut Relsen

WPS218 Public Finance,

Trade and
Development Vittorlo Corbo

WPS219 Rural Credit Reforms In LDCs Avishay Braverman

J. Luis Guasch
 PPR WorkingPaper Series

Title Author Date Contact

PS220 Capacity Buildingfor Policy Analysis:

Lessonsof Experience Samuel Paul
David Steedman
FrancisX. Sutton

PS221 Real ExchangeRate Uncertaintyand

Exports: Multi-CountryEmpirical
Evidence Ricardo J. Caballero
VittorioCorbo

PS222 Why StabilisationPolicies In Zambia

Did Not Succeed ChristopherColciough

PS223 Overvaluedand UndervaluedExchange

Rates In An EquilibriumOptimizing
Model Jose Saul Lizondo

PS224 The Economicsof the Government

Budget Constraint Stanley Fischer

PS225 TargetingAssistanceto the Poor:

A MultivariateApproachUsing House-
hold Survey Data Paul Glewwe
OussamaKanaan

PS226 Inflationand the Costs of

Stabilization: HistoricalCases,
Recent Experiencesand Policy Lessons Andres Solimano

PS227 InstitutionalReforms In Sector

AdjustmentOperations Samuel Paul

PS228 EconomicPerformanceof Developing

Countries In the 1980s Robert Lynn
F. DesmondMcCarthy

PS229 The DemographicTransitionand Saving

for Life-CycleMotives in Developing
Countries Steven B. Webb
Heidi Zia

PS230 Unemployment,Migrationand Wages

In Turkey, 1962-1985 Bent Hansen

PS231 The World Bank RevisedMinimum

StandardModel Doug Addison May 1989 J. Onwuemene-Kocha
61750