World Bank Reprint Series: Number Eighty-one
I. Adelman, M. J. D. Hopkins, S. Robinson,
G. B. Rodgers, and R. Wiry
A Comparison
of Tw7o Models
for Income
Distributlon PYlanning
Reprinted from Jon r,ial of Policy Modeling, vol. 1 (1979), pp. 37-82



A Comparison of Two Models
for Income Distribution Planning*
I. Adelman, University of Maryland
M. J. D. Hopkins, International Labor Organization
S. Robinson, World Bank
G. n. Rodgers, International Labor Organization
R. Wery, World Bank
This article compares the results from two models developed to explore the relationship
between policy instruments and income distribution in developing countries. The
BACHUE model is a long-run economic-demographic model of the Philippines and the
Adelman-Robinson model is a medium-term computable general equilibrium model of
Korea. In spite of differences in focus and design, the two models yield similar conclusions.
Althougi. the size distribution of income is quite stable and hard to change, it is possible
to achieve egalitarian development strategies with integrated, comprehensive sets of
policies. However, the distribution of income by socioeconomic groups (an extended
functional distribution) is much more sensitive to policy interventions, indicating the need
to integrate economic and political analysis of distribution policies. In any case, a general
equilibrium analysis is important since indirect effe';ts are often crucial in determining the
ultimate impact of a policy intervention.
INTRODUCTION
In recent years, the focus of attention in the development community
has shifted from preoccupation with economic growth to emphasis on
distribution. The major issues relate to the possibility for improving the
lot of the poorest groups of society in developing countries without
massive, abrupt, structural changes. The unresolved questions are: does
a feasible package of non-revolutionary policies exist, which will, in an
acceptable time frame, lead to the satisfaction of basic needs for the
majority of poor populations? If yes, what does it consist of? How hard
is it to implement and at what social, political, and economic cost?
In this light, two major international institutions-the I.L.O. and the
World Bank-have been supporting large scale modeling efforts to study
policy for income distribution and employment in a dynam;c systems
framework. The foci of the models differ: the I.L.O. World Employment
Program model focuses primarily on population and employment in a
medium to long-term time scale; The Adelman-Robinson World Bank
Address correspondence to: Sherman Robinson, World Bank, 1818 H Street, N. W.,
Washington, D. C. 20433
' The work described in this paper was supported by the I.L.O. and the World Bank.
Any opinions and conclusions presented in this paper, however, are the sole responsibility
of the authors.
Journial of Policy Modeling 1, 37-82 (1979)                               37
© Society for Policy Modeling, 1979



38                                                    1. Adelman et al.
model has a sXlort to medium-term economic planning orientation.1 In
each case the work has been country specific, BACHUE-Philippines, the
I.L.O. model, is the first of a series of country studies aimed at eventually
producing a viable policy instrument for integrating population policy
with employment and income distribution aspects of development.2 The
Adelman-Robinson model of Korea aims at developing appropriate
planning techniquies for coordinating the design of econonlic develop-
ment strategies with income distribution goals.
The paper starts by presenting the basic structure of the models in
skeletal form, As a background to policy analysis, the economnic and
demographic characteristics of the two countries concerned are also
sketched. We then turn to an abbreviated analysis of policy experiments
and development strategies. A comparison of the outcomes of analogous
experiments performed with each model shows that similar results are
obtained. The policy implications are therefore mutually reinforcing and
lend greater credence to the results obtained with each individual model,
2. THE MODELS
Each model is designed to provide a laboratory within which one can
explore the potential impact of standard economic policies and programs
intended to improve the relative and absolute incomes of the poor., The
models trace both the direct and indirect determinanits of the distribhitioil
of income. The structures of the models are set not only bv the nature of
the major economic forces and, policy instruments influencing the
distribution of income, but also 'vy their respective time frames. The
Korean model focuses on the short andl mediuin term, by contrast, the
Philippine model focuses on the medium to long run. Each model takes
a systems approach in which not only partial and direct relationiships.
but also total and indirect contributions to any given outcomne are
incorporated.
2.1. Summaries of Model Structure
2.1.1. KOREA
The Korea model works by depicting the operation of factor and
product markets with profit maximizing firms and utility maximizing
houlseh,Olds. Its distinguishing features are: (I) it is dynamic, with
I We are here reporting on the first distribution and poverty focused model supported
by the Bank, the Adelman and Robinson model of South Korea. The Bank is currentlv
supporting related modeling efforts in Brazil (by L. Taylor and F. Lysy; the article by
Cardoso and Taylor in this issue of JPM is a part of this study) and Malaysia (by M.
Alhuwalia, C. Pyatt, and F. Lysy).
2 The work is mainly financed by the UN Fund for Population Activities, Other related
models are being developed for Brazil, Kenya, and Yugoslavia.



INCOME DISTRIBUTION PLANNING MODELS                             39
imperfect intertemporal consistency; (2) it solves for incomes and prices
endogenously in both factor and product markets; (3) its solution is
based on achieving a measure of consistency by market clearing process-
es among the results of individual optimlizing behavior by a large number
of actors (1Lousellolds. firms): (4) it incorporates income distribution,
monetary phenomena, and foreign tr-ade; and (5) it allows for varving
principles of market clearing and institutional behavior.
The overall model consists of a static within-period adjustment model
linked to a dynamic intertemporal model.3 Within each period, the
degree of adjustment is constrained by the existence of capital in place
of a specific type; the immobility of the self-employed both in agriculture
and in urban productiorn: by inter-regional immobility of lahor; by
ri-idities in relative wage structures; and by governme it constraints on
firm behavior, especially in the foreign trade sector. Between periods, a
limited degree of flexibility is provided bv capital accumllLulation. labor
growth, rural-urban migration, changes in the amount of self-employ-
ment and changes in the size structure of production. Thus, although the
model is broadlv in the neoclassical tradition, it has a number of
disequilibrium, non- ne0classical features.
The model is reasonably disaggregated: it has four firm  (or, in
agricUltul-e, farm) sizes in each of 29 sectors (about 110 protdUcers); it has
15 household categories generating demiands in eaclh sector; and 6 skill
categories of labor operatimg in each of the firm sizes and sectors (about
500 wvage rates). As a result each period's solution yields over 3000
endogenous variables.
The model is quite comprehensive in the nurnber of features of the
economy which are endogenous and mutuallv consistent. Indeed, conl-
sitenc% is maintained among all the elementV, of the social accountingp
matrix, and most of its components are modeled endogenousl\.4 The
model goes explicitly from endogenously determined factor payments
and eniploynmer.t to household incomes, with savings and expenditure
decisions being modeled at the household level and factor employments
and returns at the Orm level. Accounting consistency is mainitained
mnoing: household, firm, government, and trade accounts; national
ilncone accolntll; ilnpUt-OUtput accou nts; the national product accCOuLns;
.1n1d the labor force anld the number of households.
.Stutic Part. A simplified version of the core static economic model is
sunm11mar-i,ed in Table 1, Part A. It is a compuitaLble general equilibrium
(CGE) model in that prices and supplies are avunued to adjust so as to
clear all markets. stiuiect to various constrainl s on factor mobility that
3 For a discussion of this approach to dynamic models, see Robinson (1976a).
'I Social Accounting Matrices are described in U.N. (1975), Pyatt and Thorbeeke (1976),
and Pvatt .and Roe (1978).



Table 1: Simplified Equations for Korea Model
Equation              Number of                   Definitions                               Conunents
Variables
A. Static Models
(1) x%s =(WIL  , ki, I j)             n       Xi' sectoral output                     Production Functions
r, productivity parameter               Used Cobb-Douglas functions in L and
Lt, labor input, X = 1, . m skill cattiori s  K or two-level CES functions w.
Ki capital stock                          labor aggregation function.
ji intermediate inputs,j  1,   n
(2) fji AjXsx5                      n n       Aj, input-output coefficients           Intermediate Goods Demand
(3) Pi = Pi -   A IgP                 n       Pi goods prices                         Net Prices
J                                P* net prices
(4) WA = P, EL                      n*m        WA wage of labor type A                Labor Demands
Equations are inter-dependent and must be
solved numerically for LA2.
(5)    a=     K                       n       W'   rental rate f capi                  Capital Rentals
Capital is assumed fixed within periods, so
rentals differ across sectors.
'6) L=d L-   I                        m       Lx labor demand                         Aggregate Labor Demand
(7) Ld4 -     0=                      m       T4 fixed labor supplies by skill category  Labor Market Equilibrium
(8) YA =    QA WA + Qh,,,+, WK      m + I     Yh average household income             Average Household Income
A                                    Qhx number of workers of type A in      Elements of Q are assumed fixed. There
household type h



Qh,m+l number of profit earners in        are rn + I household types.
household type h
W K average non-wage income (computed
m+1                                     from Wi K)
(9) NhH =     FL (Qt)kh             m + I     NhH number of households of type h      Nurnber of Households
kkL number of workers, category k       Based on fixed compositions of
NkL = L, k = 1.      m                    households by worker categories.
N       profit earners
(Q-')kh inverse of Q
(10) Cih = g1h(NVH, Yh, P . .. *,)  n  ((m + 1)  C1h household demand                Household Expenditures
Savings are calculated first.
(11 ) x                    ' -  d C1 T  j  + A BJd  n  Xd demand for good i          Aggregate Sectoral Demands
B Bcapital coefficients                                                     Savings are determined to pay
Zfr..xogenous investment demand by        for investment goods.
sector of destination
(12) X/-X7=                           n                                               Product Market Equilibrium
(13) P5 =  6 6,P                      I       P; average price level                  Prire Normalization
6, price index weights
B. Dynamic Model
(14) 2f--   (X e. p, WJ, W, Ke,r1,r2)  n     e superscript for "expected"            Demand for Investable Funds
r1 interest rate, organized money market  Real investment demand based on profit
r2 interest rate, unorganized money market  maximizing behavior, given expectations.
(15) Z5 = f s   . ( ' EFI LF2)        I       R:E retained earnings by sector         Supply of Investable Funds
LF supply of loanable funds to organized
money markets
LF2 supply of loanable funds to
unorganized money market
(continued)           4



Table 1: Continued.
Equation             Number of                  Definitions                             Comments
Variables
(16)    i   z-s2  =                                                               Money Market Clearing
Determine r1 and r2 by separately clearing
organized and unorganized money markets.
(17) K   jK' 7 . .. ,           s-')                      (. {time                 Capital Stock Updating
r gestation lag         Capital not distinguished by vintage.
(18) Ls =fL(t, W,..., WX)           m                                             Labor Supply Functions
Includes exogenous labor force growth by
skill category and rural-urban migration as a
function of income differentials.
(19) 7  fT()                        n                                              Technical Change
Exogenous rates of disembodied neutral
productivity increase.
List of Symbols: Roman letters without a bar denote endogenous variables in the static model. All others are parameters or excgenous variables.
Subscript i orj refers to sctors (n in all); subscript X refers to labor types (m in all); subscript h refers to household type (m + I in u' !i Variables
with a bar in the static model are updated in the dynamic model.



INCO(ME DISTRIBUTION PLANNIN( NID( )E)IiLS                            43
prevent the economy from     fully  djliuJStinig bv meains of pure market
mechanisms. The solution is calculated hb inca is of a ta tonnemenL
process which simulates inarklet helia%ioi: nto acuttal transactions take
place until the solutioin is reached. Thle outputs of thec static model are
,,actual" prodLuction, employnment. prices, wages, and income distribu-
tion for the period.
Equations ( 1 7) describe the belhavior of the priodLucing Ssectors Of the
economy and the workings of the factor markets. (iveen an initial guess
at solution prodluct prices, prodlucers determine th ir profit mud\iniiing
demnands for labor and intermlediate inputs as well as production (ECqs.
(1 4)). Sectoral capital is as.sumned immllobile witlin periods so that rentall
rates (Eq. (5)) can differ among sectors and all sectors have increasilng
short run cost curves. The sectoral demands for labor are agore.uated
(Eq. (6)) and interact with gi%en labor supplies to determine eLuIlib1ri u iii
emiplovmeint and wages (Eq. (7)).
Equations (8) -(14) sumilnariie the eqLudti0n1 Le.alinlg With the distl uhull-
tion of income and     the demiiand  for pIroucIts. Ihc factor m1;arket
eqLuations determine the diistribhuton Of 1n11C0im   to thle CL Mno0nlinicall
active population. Equations (9) and (10) dLeterminiike the :lLi111her and
average income of hOusehold,s b, dlilerenlt txN es usilng in l;i, tolln on
the worker conmposition of                'Itiisela ilds.5 Ihe model aiso gcn,iile,s the
entire size distributiotn of inct inc using. a telniLuLe thIat aOWtrcig.ics
within-group distributionls. The algorithm used is described in Robinson
(1976b).
Equation (10) generates the dendal for prodLCts, by con.sumers. We
assume that each category of household acts as a utilitv miaximlizer with
a set of resulting expenediture funlction0s.  he functins we use are based
on Houwthakker's direct aildi-hw    utility functiOn.e1  q.uanation 10 also
jmplicitlV inCludes savings functions. In the fuill imodlel, sa; ris are
L.enerated by housesholds, firms (with sepa rite behavioral equiallions)
government, and the foreign sector     all of which are omitted in the
simplifiedl system presented here.
Aggreoate demaznids for all prodtslULN are dLetCrnniiiied in Eq. (12) by
agagreg_;atii  consumer, interlmediate ndu iie'.etinlent dIem1.mnds. Invest-
ment bv sector of destination is e\i)Lel1lou  in thie Static phase an1d iS
dICterm iinL  v by a bchavioral  odlel in thle di anilii part (described
helow). In the fuill miodel, real ine-,stinen t is a1lso aifectedI by, intlAla0ti.mAr
nmech.lismlls that are omitted here.
EqLuation.s (12) and (13) give tlle condition0  for detei ililig m.iikel
clearing produlct prices. [Flu:L;tion ( 13) is thle price  t orn iA.it1ion ruile that
5 'he leciiiiqL]je used is desribed nmore fully in Adelman and Robinson (1978al.
6 For a siirveN of expenditure systems, see Birown antd Dceaton (1972).



44                                                        1. Adelman et al.
is required since Ells. i  12) reprecct a Walrasra n econoim     that call
onlv deterini niie redati% e c      le ' l tiOl .1)1pproach we tise eswentiallI
treats the various e\cei¾, demand ctqI.Lt11ilon  (EqIs. (7) andi (12)) as a set of
simultalleloIN llu irrill r 1ir luchiarmc cqua v r. s that in uNt be d irLectll solved.
Ihe aleorithii does v,.;    ciL ime app. o\ritnlmii of miiN funrictions. any
special normwinlli;atimi iiles. or an'. reci,trig of the model s50 thiNit it catni
he expres.etld as an e\t11CrcH a  )iNpolNi. BaeCd Onl otir experlience and that
of miore receilr lidlt.LIC% in tile .. me trditidlon. the ialppro.ilh seems to he
quite elliciriir .Al ld Is  i it.l1' 110t d10I1rri.Lted byv any alttern tive solution
tech ajiiresl\.
TI"he s    Iied         pilliin 'd l t'k &O Cd here i'r.d es ouit international trade
completeiv. Ilhe full model i ric ltle, two K irids of traded -oods. for
heavily pr otected iaport' or subsidul.ed C\pollK    we assmic that thle
dilolestic markets are iiiiua.rcd ft i'in the r .st of the world arid prices are
detCrmlined so as to   tear them. Ce\AiL rl as with rlon-tr.ldcd "O odIS. [-or
these goods. Ir.tdc is dc.v imirrl b' txed rehidtiollsihips to   uiPppN  or
cate-ori:us olf derailad. ()thier l iled E!OOIdS aIre &slilimed to sell loOInesti-
callv at their i.\%orld rrt ices plis az lixedi t1ariff or urs,ldy. For these goods.
the do )rimtstiC price is tiedl to the world price rid eL\prorrts  1d imports
adjust so ais tot clear thle dorirt  r: rrimii ets.
Inler'reli'lv! Mod(Tl. I11 h '(eqi Ollrtsii (of the riterteniporal miodel are
'uIninried      Ii 1Va  i I. 1P lit 13. I be,-e aIr  essential). tilhlee  d'rimi
submodlletIs d    raiI r r!ii i   i I thIeI .  i II ra  :1 rl lI .IItiOn11 of in I.I estrert: (2)
the gror. tl'r a id 'nIp': ti  of tO r le '.l. 'in' siUPlyN .antd (3) techniiological
clhan,ge aind Ce\lltionS paj%1.lutt i
Invest miellt is JIct:rn_.:ic,d hx a on k1el If the loanable fLIrLris markets.
Producers form tl        i Ler   I ti d   !0.rr   it i.Ible hI IdLi l   onl thle basIs of
expected sas> and e'.i'e' +'      r ji   e  oif ripi s I Iq. (14)). t iedit is theni
ratiOlle  eith il b.l I ,e :  an  ii t1v1"-t rate and .illlii thle ir1:ir'ket to
clear at that rate or 11K  > settim, a tai.'I t " te Of C\IT,1dii st1 of credit and
Allow i, r  tile rate of interest toi .ali inh Ill order to cleari- the oa nable fririnds
market (Eq.L (1     and l 16f). I hII         O F'rtof tile srhrnodel are the
allocation of lto.r 1iabi  ftlldllS  fi: ia 11' and sectors and an overall
iniec tiorl of vI edilt into thle CLI hitj Il  the full m.idel.. thils injection
affects the 11r11.Itirn. rate and reAl liest1i11.',1 Ill tile ltatic m11o0del. In th1is
NimIpl1iiC'l 1110deitid l It kl; .r 1I;!lis iNI' rI iiierile it. Ihe actual i11istalIed
capital st(ck for rUe hr`t Nerio0d Is .t Ir in Eq. (17) irid deplmids oni pi5t
capitall stock"s. I.prlks .., I  rl anid p.r  ni" r  wi:1th   c1!estatroIl llags.
*\i.1grerarte lab j   ,1 A !' tr.i2d to urkx'.:1 AI  .'\V"rTliOrsl0tl'. Npecilil Irae.
I iltl some  tlI Iiit'(.Iii rlI. lii i Ii   4. II I   I  V l..  t(r  r'eflect L 'el.r r'i es  in  skill
conplll|sibhtio)li.  rakll .'a iiiio .iil ii iY ..\]Cit:l\E  i .Mdeled as a funci,tion
of ruiral-rirbani incomie  flr vi.iit. x. 1   .ll uairppel limluit onl thle possible



INCOTE DISTRIBUTION PLANNING MOI)ELS                               45
annual rate of rnigratioun7 These demographic change functions are qLuite
simple. reflecting the short to inediumni-ternii focus of the inodel in
marked contrast to the BACHLUE model.
Technological change is assumed to 'ne diseniboldied, neutral, and to
grow at exogenotslv specified rates that difler bv sectors. TIhere are also
a number of paraimeters (such as, for e\a inple. imrport etnd export
coefficients) which are updated as part of the intcirtepor;1l modmlel.
2.1.2. Tiii. P1ii.APPiI Ni( ))D I X
Themodel,namedBACHUE,9 simulates the bellh\i)r in the long run
of economic and demographic aspects of developnment. The maini
features of the model are: a rellresentation of two forms of dlulalisn, both
between urban and rural areas, and within them, endogenous population
size, structure and spatial distribution; incorporation of a number of
different forms of labor supplv, use and wage determination: and
endo,enous diistribution of income to persons and lhouselidlds. TIhe
model is both dynamnic and recursive. It treats the process of gro\s th as
a disequilibrium one. characteri.ed bv lagged reactios in mnany behav-
ioral relationship,s-a process of continually adjust ng towards eclqfil ibia
which themselves chanige over time.
The model is not based on a theor\X of optiniia.tion andtl thus does niot
fall in the neoclahsical tradition. It simulates real worltl phenom11ena.M
incorporating observed relatonships. at hboth macro and nmicro-le% eN. The
unit of micro-level analysis is the 1h1OLusehld. Ihe conseuences of
houLsehlold decision n akinue in a number of areas are built into the mnodel.
including fertility, marriage, nmitratio)n. the allocation of income betweenl
saving and consumption. expenditLure patterns, and labor force partici-
patioin. This household model is in teerated into a x% der niacro-model.
which covers the relationships betwveeni den ind, (onitlipi. emplolymen
and incoines. The macro model, and the aeereu.miton of the micro-miod)(ll
incorporate consistency in the main economniic and demio)graphic ac-
counts. Population and its distribuLtion bx age, sex, education and
location are consistent over time- siniilarlk labor s!upply and labor use.
Natiinal income, produiction and inilul-M-OUtput accotinits are consistent
internallv and with each other.
T'he focus of the imiodlel is on popuilation. eninloilvmcit and incoilne
dist ribLi tioii in the in termied iazte t) long run. The maijor niodeling effrti
%as liherefote concentriated on the denimnoraphic. and labor and inicomiie
7 We have explored inigr.iion issues in a separate paper. experimeenting with variations
in the nilr.ition function. See Adelman and Robinson t19'7).
B The model is described in full in Rodgers. flopkins,. and W6rv (1977).
9 After a Colombian goddess of love, fertility and harmony between nature and man.



46                                                   1. Adelman et al.
subsvstemns. and especially on their interrelationshiips. By contrlast, somie
major econlomic variables are omitted or determined exogenously: these
include most prices and inflation, and in most runs, the determination of
aggregate investnment and outpLut. Sillce the Korean mlodel deternlines
variables such as these enrIdolI0nousl . while leaving, imlost denlographic
relationships exogenous. there is eVidlently a colsiderable deuree oif
complementaritv between the two models.
The model is disaggregate d alone a number of ditflerent dlimensions.
Some 150 basic demiiographlic categories (age. sex, location. education)
are identifi-J, and with disaggreuated labor supply. marrial,e, fertilits,
mortality, migration, school atten dallte, and household headship rates.
over 750 endog-enousL, demographic Nariables are generated. On the
economic side sectoral final demand is disacegregated into six maiajor
componenits, withl the determination of household eqlenIdilure further
disaggregated in detail. There are 8 rural and 9 urbani producing sectors.
and value added is distributed in some 40 profit or wage-earning
categories. Income distribuLioLns are, for convenienCe. L0comp.utedL in
deciles. Overall there are over 1000 endooenous ecoInomi1ic variables.
However, the number of endogelnous -variables nxty be mnisleading
because many are generated by a small number of mitriclidl equnations
(and given unlimited data and coriputing facilities. the size of the
matrices could be expainded at s ill). A better idea of tlle strucuLiall
dimensions of the imodel is gained from the fact that it can he ful[
described in about 250 heha% ioral equatiows and identities.
A simplified and summnari/ed version of the model is given in Tal-de
2.10 There are three main suhbsstenms  economic, sumnmnarized in Eqs.
(1)-(5): labor and incomie. sumnmarized in Eqs. 16) (13); and demograph-
ic, summarized in Eqs. ( 14)- ( 18). The  nomic IubsytemI is a dLemIanWLI-
based multisectoral model whichl generates sectorl. outtputs and  alules
added. The emnphasis given to demand derin es fromii the fact that demanid
variables constitute a major potential link betweei populaltion gro%% th (on
the one lhand, and employment and inC0me d]istributioln on the other.
Equations (1) and (2) show demain d as partlk determlined by household
consumptionl, in tuirn a function of population size, structure and
locationi. Exports, government xpellnditure, and prilate in%eestmlent also
contributle to demlalnd. lnvesttineit is liUbsellUe tl\ distribLuted b  sector
of des i lnation flm rouhi sectoralI accelerator funictlion,. Do mestic   h
and inmp)omts are jointly dletermmL nedi by Eq. (3). %%h1ich also I iwom prn Ites
e.xogen0 l  s%upply constraints. modele(d by im1posiIng an overall le C\c of
total oUtpuIt b)aseC'dL on planniniig- targets, or -ftinij tes of tossille giosmlt
rates. The outpuit rIo'Mth deman'11,1ded in anN sector is revised so that tle
overall constmallit is not e\ceeded, anldlI a proplomtion of tinial deml'11m nil ma
i0 For the sake of brevitv. most accounting and identitv equations have been -mined
ItoLeiher with a number of behavioral relationships of se:ondars- importance.



Table 2: Simplified Equations for BACHUE-Plilippinesi
Equation                                       Definitions                        Comments
A. Macro-Economy
(1) S[tJ, CMti = f(X[t - l.PA[C[],P[t, Q[t- 11)      Y household disposable income      Household Consumption (C) and Savings
(S)
Disaggreggated by economic sector,
rural-urban location, and income
deciles. (13.2.10)
P 4[t] number of adults per household
PC[t] number of children per household
Q relative price of sectoral output
t timea
(2) X D  ([11 -A)Y'(C + G + I + E)                   A input-output technical coefficients  Resources Demanded (X D)
matrix                             Disaggregated by economic sector
(13).
G government expenditure (usually
endogenous)
I investment (exogenous growth)
E exports (exogenous growth)
[11 identity matrix
(3) XS = (aX D + E )O                                a coefficient determined by propensity  Domestic Supply (Xs) and Imports
to import (changing as a function  Dissagregated by sector (13).
of time)
0 optional coefficient (not sectorally  Domestic food crops are independently
isaggresated) to constrain aggregate  constrained. Imports follow from the
output to an exogenous value       equating of supply and demand.
(continued)



Table 2: Continued.
Equation                                       Definitions                        Comments
(4)LQ[t]tf(X D[tI X3[-1lJ,L[t- l],aL1/at,Q,1[t - I)  L employment                       Relative Prices (Q)
Subject to the constraint that the
mean price level remzins unchanged.
Disaggregation between agricultural
and non-agricultural sectors for
comnwodity price, by 13 sectors for
output price.
Subscript refers to sector 1, domestic
food crops.
(5) V = f (X, C., C,, Xm Ls,L,,, Q)                  C;,, C, Urban, rural household     Distribution of Value Added (V)
consumption                        Separate functions for corporate and
non-corporate profit shares and wage
shares by 2 skill categories;
disaggregation by sector and location
(13 sectors and 2 locations; total 17
categories, as most sectors occur in
only I location).
Xm share of modern sectors in output
L, L,, skilled, unskilled employment
B. Labor and Incomes
(6) J[t] = f(Yt - 1], HiJi - 1, F[t - 1],            H1 distribution of household heads by  Labor Force Participation Rates (J)
EN[t - 1], L.[t - I])                          sector                              Disaggregated by age, sex, location,
marital status, education level, and
relationship to household head
(15.2.2.2.3.2).



F fertility rate
EN school enrollment
Lm modern sector employment         (multiplied by population matrix to
obtain labor supply, J*).
(7) U[t]  f(W* [t - 1], Wlt - I], 8Lm/at)              W* wage expectations (lagged function Unemployment (U)
of previous wage growth)           Disaggregated by location and in
urban areas, skill (total 3 functions).
(8) R[t] = f (W[t - 1], Wd [t - 11, LO[t - 11, R[t - 11)  W wages (and wage ratios between  Occupational MoLlity (R)
skills or sectors)                   Functions specified for mobility
between sectors, and between
employment and self-employment,
rural and urban (19 equations).
Subscripts o and d refer to sector or
occupation of origin and destination,
respectively.
(9) Lmj = f (, Rj,k)                                   H number of households              Self-employment in Modern Sectors (Lm,j)
R,k mobility between self-employment  Disaggregated by location (2) and
and wage employment (subscriptsj    modem sectors (self-employment
and k refer to self-employment and  occurs in 7 urban and/or rural
wage employment, respectively)      sectors). Value added shares are
known (eqn. 5) so incomes follow.
(10) W., L. "  f (VI jll* Un, XI.q)                   VJ unskilled w-age bill              Urban Unskilled Wages and Employment
('I LJ)
Un unskilled unemployment           Disaggregated by sector (6 modern and 2
traditional sectors).
J* total unskilled labor supply
XI - q parameters establishing fixed
relationships between sectors,
minimum wages. etc.
(continued)   4



Table 2: Continued.                                                                                                             °
Equation                                       Definitions                        Comments
(11) W, Ls = f(l, Ln, K, RJ k, Rkk)                  K capital stock                     Urban Skilled Wages and Employrment
Rk,k inter-sector labor mobility   Disaggregated by modem sectors (6).
Traditional sectors are obtained
residually.
Wages follow as the shares in value
added are known (eqn. 5).
Rural modern sector employment and
wages are obtained from a similar
function to (11) (4 sectors), and rural
traditional sector employment
residually with certain additional
equations specified to allow for inter-
sectoral relations, such as substitution
between (modem) export crops and
(traditional) domestic food crops, and
for multiple job holding.
(12) Y- = W +    WbilN                               b, sector/skill/status distribution of  Household Income Before Remittances and
non-head activities (endogenous)    Taxes (Y*)
Disaggregated by location (2) and by
sector/skill/status of household head
(9 urban, 5 rural).
N number of non-head actives er    A variance in each sector/skill/status/
household                          location category is also computed
and mapped into household income
variance. Lognormality is assumed to
obtain decile distribution.



(13) Y = f(Y*,T,ZZ2 Yz ,     )                         T income taxes                      Househiold Income After Remittances and
Taxes (1Y)
(  a measure of the extent of recent  Disaggregated by location (2) and decile
migration                           (I o).
Z2 proportion of population urban.
Y Ir* mean urban and rural
household incomes, respectkiely
C. Demographic
(14) Pttl -P[t - I ]2[t                                P population                        Population Accounting
Disaggregation by age, sex, location,
education (15.2.2.3).
U2 demographic rate matrix of which  Schooling rates derive from an
components are survival rates,      educational sub-s, %seem in which
nigration propensities, fertility rates,  enrollment and graduation rates by
schooling rates                     sex, location, and educational level
are treated as exogenous policy
variables. The other rates are
generated by behavioral functions.
()    II ) r B[t -   I BI  11) Ba' r:.l               Y        disposahle household income Lite Expectancy (D)
Disaggregation by location (2), death
rates also by age and sex ( 1 5.2).
B measure of income neqtiality
lb I       1t[r]  11  - I . IIJ:  11. B[t - 11| .       PP rural and urhan population      Migration Between Rural and L rhan
't tI p'r     11)                                                                  Areas (Al)
Dis.agrer;iliin bx azge. marital status,
sex (using "micro" migration
functions). separate functiLnN for
rural-urban and urban-rural. and for
different education levels.
icontinued;



Table 2: Continued.
Equation                                         Definitions                         Comments
117) Fft4 f{DID  - l-JItr - 11, LAt - 1],E*t - 1])     Jf female labor force participation rate  Gross Reproduction Rate (F)
Disaggregation by location, (2);
marital fertility rates, by age (7 child-
bearing age groups).
L. proportion of employment in
a-griculture
E* proportion of adults with less than
prima-y school completed
(1S) H   hP                                            H number of household heads         Household Formnation and Category
(-no. of huu%eholds)                Distribution (H)
Disaggregation by age, sex and
location (15.2.2). The second function
gives distribution by sector, skill. and
status category 1.
fH    t H. L,)
h headship propensities
a The time subscript is used if variables from more than one time period enter into the function. In equations where all variables are from the
same period, the time subscript is dropped.



INCOME DISTlRIBUTI'ION PLANNING MODELS                       53
then be left unsatisfied (implicitly through a simple monetary inflation).
Alternatively, the supply constraint can be replaced by a balance of
pax ments constraint, an aggregate production function, or omitted
altogether to obtain a demand-driveni solution. The sectoral components
of output. however, are alkvay endogenous. There is a separate con-
straint on the growth rate of labor productivity in domestic agriculture.
Excess demand or supply in relation to this constraint is used to generate
the internal terms of trade (Eq. 4). The final equation of the economic
subsystem. Eq. (5). distributes value added to depreciation and indirect
taxes through fixed coefficients, and to distributed and undistributed
corporaite profits, income from self-employment, and skilled and un-
skilled wage bills, by location and sector, using a number of different
functions designed to simulate long-term behavioral changes.
In the labor market, labor supply is determined as part of the
household decision model (Eq. 6), with household income, fertility,
sclholinii. and employment possibilities all included as explanatorv
variables. The six dimensions of disaggregation permit the incorpornltion
of quite different behavior patterns for different populationi groups. To
obtain employment and wages (Eqs. 7 11), the emploved are distin-
guished from the self-employed, and separate labor nmrkets nre specified
for "skilled" and 'unskilled," urban and rural, andi modern and tradi-
tional sectors. DIualism is represented by the tendency for excess labor
supplv to be absorhed by traditional sectors: inemploynent is mnodeled,
but is treated as more a social than an econonmic phenomenon. It is
incorporated as a function of wage expectations and outcomes, and
employment growth (Eq. 7), thus treating unemployment as a behavioral
response in the long term to labor market conditions, ratlher than as a
consequence of short term fluctuations. Occupational mobility is treated
explicitly (Eq. 8), responding mainly to clhanges in wage anid incOme
diff'erentials. Equations 9- 11 (and their rural analogues) then generate
emlp%loT1ment and self-employment -tlhrough labor allocation (Eq. 10),
labor demand functions modified by mobilitv (Eq. I1), and, for the self-
employed, implicitly through asset ownership and inheritance modified
by mobilitv (Eq. 9). Since value added shares are known, wages are also
determ-inedL l). Dnam inc adjustments in sLubsequent periods (e.g., thlrough
value added shares, labor supply, labor mobility) occur in reslponise to
diequilibia1 produced by the current period solution,
'The wages and employments computed by this system are then
aggt!regated into ho0Usehold incomes (Eq. 12), with both means anid
X axi    C0ces computed. A remittance function allows for urban-rural
income transfers (Eq. 13) and overall distributions of gross and net
household incomes can then be obtained.



54                                                  1. Adelman et al.
The demounraphic sxstem  deierml1im-s the size and(i li-LCtIIrI- Of the
population. by are0 se.so       on. edk ln-a 0i i, aind marital status, as well
as the n ti 1lin ' Oef householdIs. Thers CO }ICor  li1tii  iS a pitpiil. ti0l IcCCOUllt-
ing System  (FI q. 1 ). hilch titilizes a partlN C\i igei0U.i educational.l
subuIodlC. and bclail\iial 0 .Ii0hi,)lai) silliiUIl. iized in Eqls. (15) (17).
Life ex)ecit;a rLi. 1 fn`iic' ni otf income and iieqnt lii>. (14. 15) is miialpped
onto .w2e-5C\ Specitic death im aes. Rutral-urban migration (Eq. 16) is
dietermilined't1 -'iii l\ b a icroll-it i 1 Litilizing peroinal chiaracteristics
(age. sex, marital stzitits, edtication) .and at mazcro-funictioni bas,ed on
interrc'' it ial dill 1rencc  In HvCO a  ls,  nle distribution, anid popuilation
size. Thle fertility f'Ll ti0Wi (1q. 17) is likewise baisedk oln houLIsehold
decision tlheorV. miic,ro r.ttiu, nsucl sariables as mortality. labor force
particilpatiml. e1llo\t in1t structUre Alld i li,i)l a hi.1 ; N M ianae ratle are
also endtil,emii, as part of the h1mionseA,ld decision imiodel. flhe numbiilher
olf 11houeholds. s,iqratc '!vm niui i iacC becmiuc of the pre\ ailcn:c of
extended famiiliesi. is a ft i'twn of population size and structure (Eq. 1X).
anid howsehold beads .tare        Ivc tcd h'v .t ir skill sta,ts camteor\ as a
functionii of the distributioin of tihe whoile populatiOnll anoe, these
*;aleu,ovries.
It is app re il even fr. ': the -siinll edcipt unof' the model, that
the dlirect fe.ldKick.l bcts'. witIlli ccIn c .aad dlenumgiaplhilc ';imihles are
extellnsive. e'liv. 1-,o ttI'  from tthe vcc'mim 'ic and labor swb-sytcmi.s to
the dare.''ui'. ouh-s\Quil are those    flecctiuuc fei tilil'.. miortalitv.
rnieration and acea ^It n:: .. . ('iio; 'c!-.3 of the cinomlic and labor
Sub-s\ siein  wich arc .1 cctls  i  iivd bx demoi! 'lihc 'ariablcs
include the lex el .II1:  c rn of h'.sch' *h! .mii1  0\,s cinenwnt c inilnmp-
tion, labhrl sn'ppl\ Ill d   a,spects of the I i li arkelC. In adkltliOn'
there are a Li    of i ;.~ e. t liihks, lo ir e \ m ai '  (a) hlmisehlild sti l,ii tnrc
affcms income d.isti ibti, I.. fl'm  il f, i nl\ si/C III   and The iLIira
tlhrouglh their rel!ti i to) ilb. -r i'mt : I ).ln 'gi :iic flictors It  inllll;'lnCn
si stermi performance inllci: .ns '.iihl as (DM' Per capita. Per houslehCldl.
per nuclear household, or per anili' vonw. ilent: (c) chances over timiie in
the l)catii ' of  Lii ciii arc cii ci a ie  bx a in m\ of econf o. .c and
demlorpiaic f;ictrs.: (Id) Inc 'inc 1a:.r  n!0111 IlU[in to rural (or vice
versa), are a ectic:.l bx Ilth rehltis simze  -i;ii and rural pt plmllio1n
and  byI) ruil-   to  iiib.iiu  Hi.  a1!1'11.
2.2. D)ata, Mode(lel Fs,tinhim 11 i. %imdcl a Wilidltinl
and( Basic MlIode t)l ynamic
12.1. KIoii x
Ilhe data base is oimiisu.ul.  ,od aind plentifutl. f le pr'iiarX ' data
SOUrce.' tiseLl are:  ii ti J tuk  ,;I irt'1'111'.10-. ani d-l-tiittpiIt t,lale, a  capital-
, F r a discus,i'n  * i thLe  .i, . ,i, k dl.lNIWej  see  chultiz  1'fl4)



INCOME DISTRIBUTION PLANNING MODELS                           55
output table, a census of manufacturing, a survey of the economically
active, a population census, a survey of financial status of firms, and a
special household expenditure survey which also specified the employ-
ment status of every household member. Additional specialized data sets
employed are: farm management information, retail and wholesale trade
census, tax data, flow of funds data, and world-price survey.
The bulk of the data work consisted of reconciling the information
from various sources and years into a mutually consistent social account-
ing framework. The principle followed was to tie all the flows to the
input-output totals for 1968. reconcile them with the national accounts
for 1968. and use other sources to break down the relevant totals by firm
size, labor categories, household types, and so forth. Most of the data
refer to 1968, but some of the specialized sources were for other years
(mostly 1970 or 1972).
Most of the parameters of the model were estimated from the cross-
sectional information underlying the data base. Mostly, they were ratios
or averages scaled to yield the reconciled totals. The major econometric
effort was in the derivation of the price and income elastic expenditure
functions by each of our fifteen household categories and 29 sectors.
Attempts to estimate the production function parameters from time
series or cross-section data (or combinations of the two) yielded erratic
and implauisible results at the level of disapggregation (4 firm sizes and 29
sectors) required for the model, the production function parameters used
were therefore estimiiated directly from the 1968 factor share data, and
tested by forecasting and comparing the results with 1970 1-0 data. As
in most economy-wide planning models, the major effort lay in deriving
a consistent data base. Once this was accomplished the estimation
process itself was relativelv easy.
The model was validated by comparing the model solutions for the
endogenous variables with actual data for 1968, 1970, and 1972. The
results agree extremely well. For 1968, 95- of the more than 3000
endo-enous variables are within two percentage points of the actual
solution. For 1970 and 1972, less detailed comparisons were possible,
since less complete data sets wvere available for these years. However,
Agireement is still extremely close (within 3%)., and thiere is no evidence
of time drift. After 1972, smooth growth rates are assumed for the
exogenoms variables, in order to allow the model to follow a relatively
smooth path so as to show up more clearly the effects of policy
experimnenits. Therefore, the basic solution of the model does not
represent a forecaist. Rather it represents a reasonable extrapolation of
trends between 1968 and 1972 which is conditiornal upon the particular
choice of: export growth: import substitution: labor force growth:
expectations: and credit, tax and trade policy regimes.



56                                               1. Adelman et al.
The basic solution path assumes continuation of the outward-look-ing
strategy of export-oriented, labor-intensive industrialization and growth.
It yields increasing growth of output, rising productivity, declining
inflation rates, improving balance of payments, and deterioratinig income
distribution. The share of income accruing to the lower quintile drops
from 5.9 to 4.6%, while the share accruing to the top deciles rises from
29.7 to 33.0%. Part of the deterioration in distribution is due to
increasing variance of urban incomes, and part is due to the deteriora1-
tion of the agricultural terms of trade after import substitution possibil-
ities in grains are exhausted. As a result, after 1972 rural incomes rise less
rapidly and the gap between rural incomes increases. The absolute levels
of income of the poor continue to rise, however, and by the end of the
simulation decade the percentage of households below a fixed povertN
line declines dramatically.
2.2.2. PHILIPPINES
The sources of data for the Philippine model are national accounits, an
input-output table, several household expenditure suirey>. censllses of
population, agriculture and industry, labor force surveys, surveNs of
manufacturing, foreign trade statistics and miscellaneous data on prices,
taxation, education, and so on, mostly from government souirces. The
data base is extensive but patchy, both in quality and coverage. It was
adequate for most model-building purposes, but some limited use of
non-Philippine data also proved necessary. The model was initialized on
an input-output table for 1965, and a consistent economic and demo-
graphic accounting framework for this year was built. This involved the
organization and adjustment of data from a variety of sources. lnostl%
from 1965 or an adjacent year, following a procedure similkr to th.at
described for the Korea model.
Most of the model parameters were estimated from cross-sectionall
data. Household expenditure and savings, migration, labor supply and
marriage functionis were all estimated econometrically from household
survey data; expenditure and savings functions were based on grotulped
data, the remainder on individual observations. Fertility and mortality
functions were based on internationial cross-section. The prrOduction
system and labor demand were based on point estimates derived at the
macro level. The sectoral disaggregaton was designied to ireduice coeilh-
cient changes in the input-output system antd the labor market due to
changes in the intra-sectoral strLucture. The mapping of wages and
employment onto household inconles was based on hOuseholdl survey
data. Additiona) dynamic parameters were ilntroduced to represent
technological progress, import substitution, or occupational nliohilit\.



INCOME DISTRIBUTION PLANNING MODELS                           57
The data base for estimating such parameters was mostly mediocre, and
many of them were specified exogenously.
The conventional verification of the model is dliffiCult, since it is not
designed to track short-term fluctuations in the Philippinie economy, and
can therefore only be validated against relatively long time series. But
such time series have generally been used in m-odel estimationi, or
measure variables which are exogenous to the model HIowever, the
model appears to generate a reasonable approximationi to the long run
path underlying Philippine development since 1965. The model should
also respond to changes and generate behavior which is consistent with
the Philippines' historical experience and with development patterns
observed elsewhere. On this criterion the model behaves well: the
balance of agriculture and industry, the responsiveniess of the labor
market to economic growth. the dynamics of the distribution of income.
and the levels of and changes in fertility and migration all relate to eaclh
other and to the historical developmernt process in a coherent anid
plausible way. Thus the long-run outcomes of the model are internally
and behaviorally cotnsistent.
The reference run of the model is desinled in some sen,se as a central
estimate of Philippine development, 1lthoUgh it .should not be regarded
as a projection without also conisiderinig the conifidenice interval alrounld
it. To establish this run we took cuirrenit a1ttitULdes to deVC10elope1nt policy
and expectations of outconmes, particularly anmong Philippille planners.
and converted them into cluamtitati\e estimates of certain externaily
specified parameters (including growth rates of aggregate output and
investment demand). The outcormie is a developmnenit path on which
modern, urban emplovment increases relatively rapidl% witlh deci-easilli
wage differentials, and constrainis on food suppiy lead to rising food
prices. Mainly as a result of these two factors. inequatliN slowlv but
steadily declines. The Philippines enters the "denmogiaphiic tranisition."
but does not complete it bv the end of the century. By the year 2000 the
country's overall level of developmenit, as; measured by GDP per capita.
is comparable with some of the more developed Latin Americani
countries in the earlv 1970's, or with Spain, Greece or Japan in the early
1960's.
3. POLICY EXPERIMENTS
3.1. The Policy E,iiiroiiinie;it
3.1.1. KomIA\
I.and tenure in Korea is quite egalitariln and tlhere is n1o tenaiic% or
sh:ire cropping. However, most of the fairms are exceediinglv small-l 90(Y
are helow three acres. Korea is a very hiigh cost producer for rice, but



58                                                  1. Adelnian et al,
agricultural productivity has been rising rapidly. The basic run assumes
that by the middle of the period (1972) self-sufficiency in grains is
attained, and that, as part of the grain price stabilization policv, some
rice is subsequently exported. Nevertheless, after self-sufficiency is
reached, the agricultural terms of trade deteriorate in the basic solution
to 86% of their initial values. Therefore, while in the heginniiig ( 1968) the
poverty group is half urban and half rural, 70%, of the poor are rural bv
the terminal year of the run. The decline in terms of trade is due to both
supply and demand effects: the rising productivity increases supplie'.,
and the general urban prosperity induced by very high growth rates leads
to a substantial shift out of direct consunmlption of agricultural prodUcts
and into processed foods and urban goods and services.
Much of Korea's spectacular recent output growth, 13('. per year
between 1968 and 1972, has been based on its shift from an inward-
oriented development strategy, pursued up to 1964, to an export-oriented
one. In 1970 export industries accounted for 13% of total value added
and for 36% of manufacturing value added. While there are some
primary exports, these account for a small share of the total. The annual
real rate of growth of exports between 1964-1972 has been about 4()'r.
The basic run assumes continuation of an outward-oricented develop-
ment strategy stressing exports of labor-intensive ma.lnuf:LiCtUres.
The labor market, in Korea, has reached the "no .Lurpluw-lAhor"
turning point. Overt unemployment is quite low (4%'), mioraition to urbanl
areas high, and labor absorption in modern sector emnplo\ ment quite
rapid. In recent years, unskilled wages have been rising. The level of
education is extremely high, especially relative to GDP per capita.
Virtually all unskilled workers have completed primarv school: and the
skilled, secondary school. Human capital is therefore plentiful and its
ownership widely distributed.
The urban poor, while poor, are relatively well off when compared to
the rural poor, and become increasingly so. The rural poor are mainly
landless agricultural workers and small farmers. The urban poor dr;im'
members from many different urban groups, large proportions of sa-idl
urban groups being non-poor. Urban proverty groups are Oherefore a
difficLult target and there are significant leakages to middle ilicomlle
groups from policies directed at them.
Population policies were intitiated about 20 ',ears ago miLdl Korei
might claim to be a showplace for sucicessful family planning. In 15
years, populationi growth has declined from over 3.0', to 2.0%; dlesrite
rapidfly falling mortality. However, it is not clear what proportion of thle
decline can be attributed to faimily planning rather thlan to r.apid
modernization.



INCOME DISTRIBUTION PLANNING MODELS                             59
3.1.2. PI-11LIPPINE.S
Despite several stages of land reform, the most recent commencing in
1972, land ownership remains uneqlual in the Philippines, with extensive
tenancy, considerable regional variations in productivity, and an increas-
ing nunmber of landless agricultural laborers. Ilowever, by Asian stand-
ards, the inequality is not exceptional. FoOLdcrops (mostly rice and
maize) dominate in terms of value of output and emnploymnent, but there
is also a sizeable export crop sector, partly plantation based (notably
sugar), and partly on small holdings (coconiuts). The couniitr\ has recently
been self-sufficient in foodgrains only in years of exceptional harvests.
but export crop earnings more than offset tlhis. Over the period
1950-1970, there was an average 2' per annum rise in agricultural labor
prodLctivit%. coming mostly from extension of the cropped area in the
1950's, and from increased yields in the 196,0's. At the same time, the
internal terms of trade moved sig,nificantly towards agriculture. a trend
which continues in the reference run.
Urban development in the last 20 vears has mainly relied on heavily
protected import substit. ;on. While this has peri1litte(I a respectable
GDP growth rate averaging aroundl 6''f, it hlas halid sevei:il lndesiralble
consequences: firstly, there have been recturring balance of pa%nient."s
crises, secondlv, because controlN on imports of plroducer goods were
less severe, manufacturing has concentrated on the finishing stages,
thirdly, there has been an increasing,ly ca;pitaLl-initensive set of technolog,
choices. In addition, the system has exacerbated regional inequalities.
These problems are recognized in the Philippines, and in the 1970's there
have been some moves towards a more open pattern of development.
stressing exports of manufactures.
The employment structure of the Philipines is dLualistic. IMoLdern
sectors pay wages significantly above unskilled earning,s in "traditional"
sectors. Re,l- wages in modern urban sectors have remained more or less
constant as the urban labor force expands. However, the overall income
distribution appears to have been fairly stable. Rural-urban income
differentials have been somorewhat declining, partiCUlarly rural. hig-city
dilferen tials.
POpulatioln growdth has been consistently high over the last 20 Nears,
attaining 3-3.5'; per year duling the 1960's. flo%%ever there are indica-
tions thalt fertility is declining, and the reference run suugests that this
ws ill co1ntinLeI. EdLuCation growth in the past has been rapid, anld relative
to its level of development the Plilippine s hi. one of the highest averalge
education levels in the world. Inequality in acceess to this education still
renfains, lhowever, especiallv at secondar'. level and above, which is
laryel\ private.



60                                                     I. Adelman et al.
3.2. Experiments
There are two major ways in which policy chaniges can be simulated
in the models: (i) by exogenously modifying variables which are or mliglht
be directly under government control-policy instruments; and (ii) by
exogenously modifying variables which, althoLIgh they cannot be directl\
manipulated by governments, can be affected by. and therefore repres"ent
the use of, policy instruments not in the model. In both cases it is the
ultimate impact on certain target variables, which are not themselves
exogenously manipulated, which is of interest, In general, the Philippines
model has more policy instruments available in the demographic sector,
the Korean model in the economic sector, and this is reflected in the wav
policy changes are simulated. Furthermore, the magnitudes of experi-
mental changes are generally greater, and their introduction more
abrupt, in the Korean model than in the Philippine model. Larger
changes are required in a short-term model to generate outcomes
comparable in magnitude with the long-term results. The detailed
experimental designs are also, at times, not strictly comparable. Despite
the differences in the economies, model structures, model time-frames
and ways in which the policies are modeled, the results of analogous
interventions in the two model economies are quite similar.'2
3.2.1. AGRI( ULT.RLRAt. PRODLCTIVITY
(i) Philippines. Agricultural labor productivity growth is raised from
three to four percent per year.
This change, at face value favoring agriculture, has a serious adverse
effect on rural incomes. In year 2513 the terms of trade have fallen by
29% relative to the reference run, rural wages by 15`?, and rural incomnes
by over 10%; the effects on rural areas would be even more serioLus were
there not significantly higher out-migration and increased remiittances
from urban areas. Demand for agricultural output rises, but not enough
to compensate for the decline in the terms of trade. The negative effects
on rural incomes are reinforced by a decline in demand from rural areas
for rural nonagricultural output, and the share of rural areas in total
output, even in constant prices, declines slightly. Rural income dii tribu-
tion worsens since it is the lower income food producers who are most
affected, Since the average incomne in the economy as a whole has
12 In the discussion which follows, percentage changes quoted relate to the ditTerence
between the experiment in question and the reference run of the model corncerncd, except
where otherwise stated. Thus if migration is "five % up in the eighth year," this implies that
it is five % higher in the eighth year than the value for the eighth year in the reference run.
If there is no reference to a time, the figures quoted refer to the terminal year--i.e., the
ninth year for the Korean model and the 25th year for the Philippines.
13 For year ten the results are qualitatively similar but smaller in magniLude.



INCOME DISTI'RIBUTION PLANNING MODELS                            61
changed little there are conmpensating gains in urban areas, with the
urban poor doing naruinall\ better than the urban rich, despite the
influx of unskilled migrants. The main demiographic side-effects are
higher fertility and mortalitv in rural areas, and lower in urban areas.
(ii) Korea. Total agrictllturazl produictivitv growth is raised by 2.5
percentage points per year. A nmarketing boar-d purLCha1.Ces and exports
cereals, up to 15'. of production. in an attempt to stabilize prices.
The experiment causes the relative and absotLute position of farmers to
dleterioralte. AgricultUral output is 20%; higher in the last period, and
since deemand has not increased as much, despite the operation of the
marketing board, the aLgriculturall terms of trade Ultilately fall to 73%e of
the basic run, The share of the agricUltural groups in the poor therefore
rises and, within the agricultural sector, the larger farmers, who have
more flexibility in adjusting the composition of their production, gzain
relative to the small farmers. The urban groups gain, both relatively and
in absolute terms. Skilled workers gain most and capitalists and govern-
ment wvorkers gain least in percent terms. Nevertheless, the equit) of the
overall distribution deteriorates, since the unfavorable rural efTects
dominate, National income is 5i'. higher, but the increases in total
income and product are not shared bv the poor. While the overall miiean11
income in the last period is 10", higher than in the basic run, the miean
incomes of the bottom two deciles remain virtuallv unclhanged. In this
experiment, there is no trickle-down at all!
(iii) Comment. The experiments deemonstrate the need to take both
demand and supply into account in settin- up coherent policies for
aLriculture. For increased agricultural prodUCtiviity to benefit rural areas;
their oUutuLt must have a market. Since the price elasticity of demand for
food is> ratlher low, this would imply at least one of the following:
substituting for agricultural imiiports: redistribUting inicomie to lower
income groups with higher demand for food, price supports, implement-
ed through government food purchases: or increasing agriCU]wral ex-
ports.
The results from both mo0dels emphasize the zane robust Co1nClusioll:
after import SubstitLutioll possibilities in agricultuie have been exhausted,
anx policy that increases agricultural output will tend to worsien the
economic situiationi of rural households, other things being equal. Tven
when, as in the Korean experiment, specific policies are used to try to
maintaini the agricultural terms of trade, it is nonetheless verv difficult to
keep the terms of trade from   deteriorating. And as the Philippine
experiment shows, the longer the time hoori70h considered, the worse the
deteriorration.



62                                                  1. Adelman et al.
3.2.2. RURAI PL 131.1C WORKS
(i) Korea. This example is simulated by increasing government expend-
itures on construction by 50%7(,. Seventy-five percent of the value added
thus generated leads to increased demand for agricultural laborers and
small farmers. Fifteen percent of value added of self-employed in
manufacturing and trade services is also transferred to these rural
groups.
The program in the initial year, represents an increase in labor
demand of about 2'% of the labor supplied by agricultural landless labor,
or .57t of landless labor plus family workers in the smallest two farm
sizes. Suprisingly. the package is somewhat deflationary (the inflation
rate is one percentage point lower) since it was financed out of increased
taxes, and also since overall production increased somewhat due to
higher labor utilization.
In year 5, the distributional effects are as expected: urban groups,
especially the self-employed (part of whose business receipts are trans-
ferred to rural households), are worse off and rural households are better
off. The primary beneficiaries are agriCLultural labor, whose real invonles
are 63%7et higher, and the smallest two farm sizes, whose real incomes are
up 19% and 8', respectively. There is also a tricklle-upl to large farmers
whose incomes are up 4%,r due to increased food consumption. By year
nine, however, the income effects for medium and large farms are badly
eroded because shifts in demand turn the terms of trade against
agriculture. The income gains of agricultural labor and small farmers are
actually accelerated. but there also is a substantial real income transfer
to urban workers.
The effects on the incomes of the bottom    two deciles are quite
favorable. By year 9 the poorest decile's meani real income is 30e.% higher
and the next decile's is 17'. higher. The absolute improvement in the
incomes of the poorest decreases the degree of poverty very substantial-
ly-bv year 9, the share of households below the poverty line is reduced
by one third.
(ii) Philippines. This example is simulated by increasing government
demand for rural construction labor. laruel\ in direct labor intensix e
w(orks. The program adds 12 13'% to government expenditure, and grows
by 5e. per year from year I to year 10 when it ceases,
The inmpact effect of the programi is to increase rural construction.
employment by about 2.5%, of the rural labor force, Over time, it tendLls
to raise rural construction %vages. but also generates substantial spill-over
effects whiclh srpmad the benelits to other rural sectors. Most of the spill-
over conies from a relative rise in food prices, (lue firstly to reductionis in
aoricUltural labor sul,P1N. and secondly to increases in demand for food



INCOME DISTRIBUTION PLANNING MODELS                               63
consequent on rises in low rural incomes. The reduced labor sUpply to
,tg iculture occurs despite reduced out-migration and a small increase in
rural labor force participation rates for non-houselhold heads. The terms
eof trade effects lead to a small de-rea.se in urban wages (mostly 3 to 4`%),
and reduced urban-rural wage differentials lead to a significant dlecline
in migration. Thus, rural public works programs help to resolve urban
employment problems through decreases in inigration. and to raise all
rural incomes. However, the direct effects of public works are secondary
in this; it is the spill-over effects which are important, especiallv after
sufficient time has passed for rnigra:ion feedbacks to have cumulated.
Even by year 25, some 15 years after the end of the program, the after-
effects are still visible. Rural construction emnployrment is still higher than
in the reference run, and the terms of trade have remained altered in
favor of rural areas. Overall, the program is somewhat, but not massively
egalitarian-the Gini coefficient declines by around one percent.
These conclusions assume that the public works do not affect agricul-
tural productivity. If we relax this assumption. and permit agricultural
productivity growth to rise, the effect is at first to diminish and finally to
entirely erode the gains in rural incomes, because the terms of trade
eventually decline for reasons similar to those discussed in Section 3.2. 1.
(iii) Comment. The two models produce results which are in broad
agreement on essentials-- the public works program generates a signifi-
cant decrease in rural poverty in both experiments. However, this
outcome is generated by different processes in the two models. In the
Philippine case, the main factors are increasing terms of trade, reduced
migration and a withdrawl of labor from agricultural production:
moreover, entrepreneurs and self-emploved in the construction sector do
relatively well. The presence of a rural labor "'surplus" contributies to the
success of the program. In the Korean model, the main elements are an
inelastic demand for agricultural labor, increased aggregate output, and
terms of trade which first rise but eventually decline. Excess labor supply
is not a factor here, highlighting the difference in overall level of
development between Korea and the Philippines.
3.2.3. EXPORT PROMOTION
(i) Korea. Labor intensive and export industries promoted by (a)
subsidizing their output prices, (b) lowering their interest rates, :an1d. (c)
increasing their ex.port targets. The sectors involved comprise 28'; of
total initial production.
The strategN inmproves the balance of payments and, at the same time,
it improves the distribution of income as well. Initially, the policy
packane clearly has some diifferenlLial impact on the outpfUt of the



64                                                  1. Adelman et al.
subsidized sectors. However, from year five on, there is no real diff'erence
in the impact on production in non-food s,uh,idized versus non-subsi-
dized sectors: the linkages through intermediate demands have spread
the effect of the program through the rest of the econonm. The initial
impact of the package is highly inflatiolnary but the inflationarv bump
dissipates quickly as aggregate output increases, and by vear three the
net effect is deflationary compared to the basic run.,
There is an increase in rural-urban migration which lowers agricultural
production at the same time as urban output, both rnallnUfLcturing and
non-manufacturing, increases. The terms of trade therefore move strong-
ly in favor of agriculture despite a relative decrease in overall demanid
for unprocessed agricultural output. The net effect is a windfall to rural
households and an income deterioration for urbani households. By year
nine, rural incomes are 40 to 52% above those in the basic run, withl the
rich farmers gaining relatively. The inconmes of capitalists andi self-
employed in traditional sectors decline relatively much less. Even thoiughi
the distribution. within both the rtural and urban sectors deteriorate, the
overall distribution improves because the rural-urbani mean income gap
narrows and migration shifts more 1101u.ellolds into the hiiher income
urban sector. The Gini coefficient is substantially inipro%ed, bhut the
incomes of the poorest two deciles and the percellt of hol0Ueholds in
poverty remains virtUally unchanged. Ihe ruLral poor are th1usE replaced
by the urban poor: their percent.ie of the poverty group more than
doubles. Thus, promotion of libor intermive industrv is in fact one of the
most effective ways of benefitting farmers.
(ii) Philippines. Manufacturing sector exports are raised oradul.llly so as
to be roughly double reference run levels hb year 25. Sectors involved
comprise 16', of initial total prodLuCtionl. Two alternatves are e\plored:
firstly, output growth is left unchanged at reference run levels, on the
assumlption that supply constraints dominate: and secondl%, output and
investment growth rates are both r-aised by one percentage point to
reflect a relief of demand constraints.
In the first alternative, total exports rise by 12% in year ten and 36%
in year 25. and the balance of pa\lllellts is considerahl\ improved.
However, the increase in manufacturing export, is partly at the expense
of exports from   other sectors. Investment and private and public
consumption all declinie to permit the allocation of es"ources to exports
and to the capital outflow implicit in the halance of payments surplLus.
Secondarv sectors increase their share in output b\ some three percent-
age points, at the expenise of both primarv and tertiary sectors.
The shift towards secoi darv sectors vt'duces rural value added bv
seven %? in year 25. leads to a decline in the agricultural ternis of trade



INCOME DISTRIBUTION PLANNING MODELS                             65
and eventually to a slight relative decline in rural incomes. Declining
demand for exported primary products leads to a reduction in the size of
the relatively privileged modern sectors in rural areas, a reduction in
their wage differential over traditional agriculture and reduced inequali-
ty. The self-employed have the largest fall in earnings. In urban areas.
increased manufacturing sector employment leads to reduced dualism
and less inequality. The improvement in equality within each region
leads to an improvement in the overall Gini coefficient, but this mostly
reflects gains above the lowest decile since the rural poor do not benefit.
In the second run it is assumed that the balance of trade surplus
relieves a constraint on output growth and the latter is raised by one
percent per annum. Incomes increase by a quarter, on average, in year
25 and the distributional pattern also changes considerably. Increased
demand for agricultural output rapidly drives up the agricultural terms
of trade. This raises rural incomes relative to urban and in fact
concentrates the gains in incomes almost entirely in rural areas. As a
consequence, income distribution improves significantly, with the overall
Gini coefficient down by seven % in year 25.
(iii) Comment. The overall inpact of the types of export promotion
explored here tends to be egalitarian in both models, though the results
are somewhat different. In the Philippine model, within-region move-
ments towards equality dominate if the growth rate remains ulnchanged
because of intersectoral shifts, With faster growth, gains in equality arise
mostly out of higher relative food prices. In the Korean model, the latter
effect is generated by endogenous output response; however, within-
region inequality increases. We should in any case note that with a large
balance of payments surplus and rising food prices, there would be
considerable political pressure for food imports, so that the extremes in
the distributional changes generated in these experiments are unlikely to
be reached in practice.
It is interesting to note that import substitution experiments, not
reported here, tend to have the opposite effects, with worsening rural
poverty and increasing inequality in both models.14
3.2.4. MIfIGRATION
(i) Philippines. Mfigration between rural and urban areas is exogelnouslv
set at zero.
This run indicates that migration plays a significanlt equilibrating role.
All aspects of demographic structure, employment, and income distribu-
14 Questions of import substitution are explored in a separate paper for the Korean
model, Adelman and Robinson (1978b), and in Rodgers, Hopkins and Wery (1977) for
BACH U,E-Philippines.



66                                                   1. Adelman et al.
tion are greatly modified. By contrast, changes in economic structure are
quite limited; there is a small shift in output towards primary sectors, a
slight decline in imports, and a small increase in government expendi-
tures.
In the 25th year, urban population is down 22%., not only because of
the absence of migration, but also because of a decline in fertility anid an
increase in life expectancy. Symmetrically, rural populationi is highier.
Total population is marginally higher and overall fertility has been
raised. The changes in fertilitv arise mainly out of mortality chlanages, due
in turn to changes in income levels and distribution. However, indirect
feedbacks on labor force participation, education, and the proportion of
the population in agriculture also contribute.
Urban employment falls, particularlk in traditional sectors, while rural
employment rises, again particularlv in traditional sectors. Thus rural
dualism increases, while urban dualism is reduced. There is a considera-
ble fall in the agricultural terms of trade because of the change.> in
production. As a result, urban wages rise by from 20 to 40;)', while rural
wages fall by about 25'. In urban areas, it is the unskilled who gain
most, since the bulk of the migrant floiw is unedlucated. In monsequLLeCe.
within urban areas the effect is maruinally egalitarian. In rural areas the
reverse is true. But much more important is the inter-rceional distribl-
tion effect, with income inequality risinp. Ihe Cini coellicicnlt is up by
6'. which is a large deterioration by the standards of most other
experiments.
(ii) Korea. M-ligration between rural and urban areas is exogenllUsl set
at zero.
The experiment shows, not surprising,h. that nmigration, up to a point,
is important in equalizing distribution by redLuCilln regiona.l income anid
wage disparities. When migration is stopped. urban intcomies rise by
around 10-15'%, especially for uinskilled and skilled workers and appren-
tices, and rural incomes fall by about 2'('<, especially for aariCultura'
labor. Terms of trade turn against farmers (they are 15% lower), as the
urban demand for food is less, the supply of urban inanufacturinaL anid
serv ices is reduced. and the supply of agricLiltural output is illcreased. All
these effects are due directlv to the migratlion clianLe. TIhe fall in terms
of trade has an eqluaflizing efTect upon rural incomie,s, as large fariners
lose relatively more than small farmners. But, since the rural-urban
income gap rises and more people are now in the lowest income sector.
the overall effect is to increase inequalitN su'tantiallv (tthe (ui coelli-
cient rises 7',;) and increase a1solutC povertv.
(iii) C'ommew. The results of both experimenits are consistent: they
emphasize the importance of econornicallk induced population move-



INCOME E`.TRIBUTION PLANNING MODELS                               67
ments for poverty reduction and income equalization. This importance
also comes out in the other experiments which indicate that, not only is
migration the most significant demographic variable in each model, it is
also one of the most significant of all variables in both models. Migration
is moreover the only demographic variable to have important effects
within a decade.
It would, of course, be possible to overdo inigraLtion and thereby
generate greater inequality: up to the point where the incomes of
migrants and of would-be migrants are equalized, migration reduces
inequality and absolute poverty; thereafter, it increases it.
3.2.5. POPULATION GROWTH
(i) Philippines. Addtional successful family planning program increas-
ing fertility decline by one percent per year over years one to ten; life
expectancy raised by health programs.; education growth considerably
increased; female labor force participation increased; output growth is
assumed unchanged.15
This program generates a large reduction in fertility-from a crude
birth rate of 40 to 26.5 in ten years. If this occurred it would be among
the most rapid fertility declines witnessed anywhere, and thus this run
probably represents the maximum outcome of a population control
program. Despite this massive effort and the pro-natalist bias involved in
the assumed lack of correlation between population change and output
change, the outcomes, though favorable, are not startlingly large even by
year 25. A 24% decline in fertility has generated a 19% decline in the
crude birth rate, 9% in population, and a 5% increase in mean income
per adult equivalent. The use of the latter variable to measure welfare
assumes that only unwanted births are eliminated. If the suppc)sed
fertility decline cannot be based on an assumption of excess of actual
over desired births, a measure of welfare such as income per nuclear
family is superior,16 and this measure does not start to decline sienifi-
cantly until after year 30. Thus the overall welfare gains, even in the long
term, are not large.
Distributional effects are mixed. The reduction in population growth
generates a shift in demand away from food, leading to a declinle in the
agricultural terms of trade and to reduced rural oUtpuLt. As a coise-
quence of this, rural wages decline, Urban wages do not rise, on aXeraLge,
1S In fact, there is considerable evidence that population growth and output growth are
positively related, mostly in forms not amenable to incorporation in a quantitahive model.
See, for instance, correlations computed in UN (1973), p. 535. Questions of endogeiiollu
fertility response to economic policy are discussed in more demail in another paper (Rodgers
and Wery (1977)).
16 See Blandy (1974) for a statement of the argument underlvirkg this assertion.



68                                                   1. Adelman et al.
because of the increased female labor supply. However, urban incomes
per household and per capita are up, and rural are dowII, and overall
inequality is increased.
This is true of year ten, and it is still largely true of year 25. But the
long run dynamics are governed by two factors, niizrationi and-
eventually-a diminished supply of new labor force entrants. As a result
of higher education levels, rural to urban inigrationl increase,s, cointrolling
agricultural output supply and preventing the agricultural terros of trade
from declining precipitously. After year 15, this effect is auglnenited by a
relative decline in rural labor supply. arising out of earlier reduced
fertility. This eventually (from around year 30 onwards) turnis the terms
of trade in favor of agriculture anid generates increasingly egalitarian
outcomes.
Within urban areas, the increased supply of educated labor reduces
wage differentials, and promotes urban equality. Htowever, because of
the increased female labor force participationi. traditionial sector labor
absorption is increased rather than reduced, and modern-traditional
wage differentials widen.
While the directions of these effects are of interest, tlheir mn.ynlitudles
are quite small. All changes in the Gini coetlicienit are less tlhain two per
cent until after year 30, and even household incoomes in the tolp and
bottom urban and rural deciles vary within a range of +2 to -5 perceint
up to year 25.
(ii) Korea. The aggregate annual rates of growth of the rurazl and Lurban
labor forces are decreased by one percentage point per vear. 'IThe
assumed change is extremely large and the implied population programn
very effective by international standards.
The effects are long run: they are barelv noticeahle by vear five and
only start being felt by year nine.
The results of implementing a natiommxide policy to limit pO)lat,1io0n
growth are bad for both distribution and absolut;le p    ertN. With
balanced change in population growth, the very differenit factor propor-
tions and techniology in agricultural and nonagricultural sectors cause
shifts in the structure of productioni. Rura;l-LuI-rban migrationl and the
investment miiodlel provide additional mecl rnliniiis for quantity aILIilst-
ments by factors of production to chalnoe, in prodIuclion sLultltire, but
price and wage adjustment occur as well. With low populatiOll growth.
the OU tpUt of both the agriCultural and nonm-auricul tural sectors is lomer.
There is an apparent elasticity of 0.2 for agoregate Output glrowtl with
respect to poPuiation growtlh and an elalctict\ of 0.6 for the aunricultur.lI
sectOrs. Lower rates of populaktioni gowth detreise thc h;hre of ioriciil-
ture both in total prodluctioll an12d in 10tt11 tiC1SuI II)ti0Io demCiri nd. The



INCOME DISTRIBUTION PLANNING MODELS                           69
decrease in the share of agriculture in demand, however, is far greater
than the decrease in the share of production, resulting in a dramatic
deterioration in the agricultural terms of trade. In year 9, the terms of
trade are 50 for the low growth experiment as compared to 86 for the
basic run.
The changes in the structure of output and demand also reflect
differences in migration rates-the low growth experiment induced much
more migration, with total migration 41% higher over the entire period.
The higher migration with lower population growth was due to "push"
factors, namely, a widening gap between rural and urban incomes. This
increased migration was not, however, enough to offset the declining
terms of trade in the experiment. By year nine, the urban-rural income
ratio is 3.56 for the experiment compared to 1.79 for the basic run. The
rapid decline in the terms of trade causes agricultural incomes to fall
faster than can be offset by migration. In terms of the relative income
distribution, more rapid labor growth leads to a more equal basic run
and slow growth experiments respectively.
(iii) Comment. Both models yield consistent, and disappointing, results
in terms of absolute and relative poverty reduction. Botth imply that
these demographic programs are primarily upper middle class programs,
in the short to middle run. It should also be noted that the family
planning programs simulated are spectacularly effective by international
standards and the models assume that the programs are implemented
without imposing substantial burdens on the government budgets-
thereby overstating their eventual beneficial consequences.
3.2.6. GLOBAL STRATEGIES
While a number of the experiments described above consist of
packages of complementary policies, none of them constitute coherent
approaches to the development process as a whole. A number of such
coherent strategies have been run with both the Korean and Philippine
models. Since these strategies were devised and tested independently,
they are r. ,, directly comiiparable, but they explore broadly analogous
development paths. It is therefore of interest to briefly describe the
content of selected strategies and their main results, especially since in a
number of ways they sum up the outcomes of the experiments reported
so far.
(i) Philippines: (a) Comprehensive employment strategy mission.  In 1973
an I.L.O. mission to the Philippines, led by Gustav Ranis, recomnniended
a package of policy measures intended to take the country along a path
leading to full employment, equity, and growth, without major structural



70                                                   1. Adelman et al.
changes.17 The package included: a rural prograim involving public
works, land reform, rural industry, and expa;ndled food pIrotdLcCtion; an
urban program promoting labor-intensive manufactured exports and
small-scale industry, import liberalizationi and a limited anti-poverty
program. It was hoped that this strateov would aidd one or more
percentage points to the output growth rate.
The simulation of this strateg\ suggests that it might be quite
sLuccessful, assuming full implementation of all its components. The main
result is a relative redistribution from  the urban middle and higher
classes to the urh;ban poor and the rural popuLlation. Elven if a higher
growth rate is not achieved. urban dualism anid urban wage differentials
are reduced, while a rise in relative food prices (of 30r' in year 25) leads
to substantial rural gains. If faster output growth is attained rural gains
are about twice as large and urban areas benefit too. Only the richest in
urban areas have incomes lower than in the reference run.
The effect on the overall Gini coefficient is substantial -- 14- 15i; down
with unchanged growth, 16-17'e with faster growth. Hlowever, the overall
package includes a probablv implalSibl% 1 irge program of transfers and
progressive taxes. Its exclusion lowers the gains in the (Gini coefficient to
ten percent. Given the size of the changees iiirleiienlteLl and the time
lapse required, this seems a rather small return.
(ii) Philippines: (b) Egalilrii .rtriegv This strategy overlaps with that
described above. but adds some new elements. In particul:ar, it ilnCludLe;
gradual nationalization of modern sectors. and higlh taxation on remain-
ing private corporations: a conisiderahNl larger program of rural public
works: wage subsidies in both rural and urban modern sectors: and
increased migration. Increased labor intensit,. land reform, faster agri-
cultural growth. and a taxtransfer programii are similar to those in the
previous strategy. l8
There are two essential differences from the OLutconie ( of the previous
strategy. Firstly, the nationalization and wage suLbsidlies lead to consider-
ably greater equality within urban areas. Ihe incomes of the richest
capitali.nts fall by half, and while waes in mnodern uirbarn sectors also fall
the decline is much less. Secondly% the gainls in rural %%ages aind incomes
are augmiieitied, due to increas,ed out-migraintio, to Frea ter AhX.lorption of
labor in public xk)rks. and to shiifts in dielmlalndl to%%,irds food becauise of
the eialitaria n nature of the changes   npleenicnted and their rural
orientation. The :igriculttri'- terms of trade rise and a jeutceLl rural
Population llhares the i1ncreaiseTd i ncomne fromn acricuHt tiWe. D)eniourapli-
17 See 11.0 (1974).
18 See H,topkins, R)dgerN and '  (19701 for a nondetalled ana!vs of fe;Li arian"
stralegies with BACHUE-Philippines.



INCOME DISTRIBLUTION PLANNING MODELS                            71
cally. the egalitarian strategy reduces the birtlh rate by 8', by year 10,
and by 10% by year 25, with the decline concentrated in rural areas. By
year 25, the demographlic effects are beginning to show uip in overall
measures of welfare. However, as in the case of the demographic
experiments reported above, the income effects of the demlographic
changes are still small.
Thus, the egalitarian strategy improves upon the distributional benefits
from the comprehensive employment mission strategy. Although subject
to the same qualification about the tax and transfer component, it
reduces the overall Gini in year 25 from 0.46 to 0.35. However, it
produces little or no benefit for the urban poor and requires a radical
restructuring of the ownership of capital.
The comparison between strategies is also sensitive to assumrptions
about the rate of growth. If the comprehensive imission strategx achieves
a higher rate of growth, as intended, it would effect a greater reduction
in absolute poverty than in the egalitarian strategy. even by year 10,
although the latter would still be clearly ahead in the reduction in
relative poverty. Evidently, in the egalitariani strategy, the sqlueezed
private sector cannot provide the same output growth as under progres-
sive capitalism. If the egalitarian strategy is to come out on top in both
absolute and relative terms, it needs to be implemented in an economiiic
system where the dynamic of development is pr(ovided by the state.
(iii) Korea. Two large, across-the-board policy packages were consid-
ered: a "market socialism" package (Package A) and a "reform capital-
ism" package (Package B). Package A was intended to represent a
package of policies similar to those frequently adopted in nionlconm iimunist
socialist countries. It emplhasizes land reform, n.ationaliz,ation of large-
scale industry; import substitution; programs of social developmllent
(consisting of COnSLuption1 transfers to the poor, large educational
investment and a very successful family limitation program); urban and
rural public works; rural cooperatives, and industrial decentralization.
Package B, on the other hand, was intended to illustrate what a
capitalist economy cotuld do if it were SUffiCientlV .-notivated by the goal
of poverty allev,iation to undertake significant alterations in the struCturle
of its society. This reform capitalism package includes the samiie pro-
grams of social development, urban and rural public works, rural
cooperatives, and indUstrial decentralization that appear in Packaep A.
However, instead of an import substitutiorn strategy, there is an elllplasis
on prorn)tion of labor-intensive exports. And, in preference to landi
reform., Packlaue B introdLuces rural nmarlketinu hoai-ds atind imlprovemlents
in agricultural productivity. There is, of course, no analogue to Package
A's nationalization program.



72                                                  1. Adelman et al.
Both packages A and B were effective. In both, by the end of the
decade, the seven lowest deciles of the society were all better off in
absolute terms than they were withoLut any polic packages at all. The
trade-offs between the two approaches;   are familiar. The market socialist
package leads to better and improving d istribution blit deteriorating
output performance relative to the Lu1nper1tLurbedl time path of the ecollo-
my. The reform capitalism package shownvs evidence of deterioration in
the distributional gains over time aned of acceleration of production
growth. By the end of the decade. ever\onc (including tlhe lowest decile)
is better off in absolute termis with the reform capitalist package than
with the market socialist one. I lowe% er, thie seven lowest inicomie deciles
are still better off under the s,ocialist package th;an in the basic dynamic
solution. The relative and absolute income implications of the two
strategies will diverge even more over 25 vears.
The reason for the lower growth rate in the socialist package is not
hard to find. First, agricultural prodL1tiC ity iS hilhsiler in the capitalist
experiment and agricultural output tlhus riscs. Second. productivity
growth in the large scale manufacturinig sectors (which initially account
for over half of total mrnufalctLurinlg olltptlt) is aSsum1led to be lower,
Third, under the socialist package, there is an. increase in ,overnment
employment and a small withdrawal of labor from the private labor
force. Finally, there is significantly more in' estnuent under the capitalist
package. The net result of these etTects is a significant difference in total
production. In year 9, the socialist packlage yields 14'; less output than
the capitalist package and 3.2'; less than in the basic run.
Not only are there differences in aggregates between the two experi-
ments but there are major structural difllelrences as well. TIhe socialist
package results in decreases in the share of the ecntomic;ll% largest firms
due to nationalization, with the relative gain going to the next two firm
sizes, while the capitalist package leaves the Structure of production by
firm sizes about the same as in the basic run. The capitalist package
subsidizes labor intensive IndsLLNtrics (of all sizes) while the socialist
package does not. T'he result is that, in Near 9, the share of labor
intensive industry is about 3 pcrcent.ige poinlts higlher in tlle caplitalist
package thani in the socialist one. The share of producer goods rises
slightly in the socialist package, undoubtedly because of the imlport
substitUtion policy. h'lie suipplv of COInsIl.nier goods, even e\cluding food,
is also substantiallv less.
Socially, the dillfCrences between the pick.mges are striking. In the
socialist pack,ae, capitalists and richi farmer\ 1,Ose aInd the 1more highly
skilled professions  especiall\ whtc collar and elngineers  g;ain. LJnder
the socialislt package the elites (tlhe upper 5" ) consi%st of bureaucrats and



INCOME DISTRIBUTION PLANNING MODELS                            73
technocrats versus capitalists ancd rich farmers under the capitalist
package, in spite of the fact that the two packages havle the same
educational program. In both experiments, rural households gain com-
pared to the basic run becaiLIsi of the improvement in the terms of trade,
but the rich farnmers gain relatively mucli more under the capitalist
package. The capitalist packa-e is also significantlv mnole faV0oraLblc to
ordinary workers than is the socialist package.
The differences between the two packages imply vast diffierences in
social stratification and in the complexion of society and the polity. The
differences are familiar, but it is very interesting to see them emerge as a
consequence merely of specific ditTerences in economic policy choices
and in economic development straitegies.
(iv) Comment. Ultimately, our global strategy analyses do not lead to
unambiguous choice between the two alternative types of ideologies
considered here. The analyses highlight the differences between them
both in economic and social terms, point out the magnitudes involved in
the trade-offs represented, and give indications Of implicit fuLtire trends.
It appears that, fundamientall%, a choice between these two approaches
must rest on the likelihood that anti-povert  programs of the order of
magnitude represented in the reform  caLpitali.Nt and comprehensive
mission experiments are possible under the capitalist mode: and whether
oir not a socialist or ega,litarialn strategy can compete with the capitalist
approach in terms of aggrevate output growth. Of course we haLL e not
exhausted the ideoloizical or practical possihilities. Further e\perinmenta-
tion might produce a strategy which can inmpro%ve on both the alterna-
tives investigated here.
4. DISCUSSION OF RESUILTS
The sirnilarity in results of atnalogous policv experiments, despite large
differenoes in model focus and nmodel strU :cure, and despite economic
and social ditTerejices in the economies heing imiodeled. is quite striking.
This leads one to ask wh\ . Possihle reasons aire: (I) pure coincidence;
(2) the model structures are broadly the same: (3) the cconornieo are
similar; or (4) while both models captutre a different reality in a differenit
manner. the strength of the efTects transcend,s these dJilVerenlceN. We
believe that the last reason is correct.
In Section 2 each niodel was ldescribed. Table 3 presents a shoirt
comparison of the salient clhracteristics of both modlels. In terms of
broad characteristics, there are a numher of similairities between the
BACHUE and Korea models that arise from the falct that hoth models
seek to "explain" the overall distribution of inc(,me over time. Thulils ileN



Table 3: Comparison of Model Features                                                                                            4
Features of the Model                 Korea                     BACHUE                              Comments
(1) Expenditure functions     price responsive              limited price response  Korea: separate demand systems for 15
no demographic variables      demographic variables      household categories.
BACHUE: two consumer categories.
(2) Investment model          behavioral                    sectoral accelerator    Korea: model of expectations and loanable
funds market.
BACH U L. capital output ratios applied to
sectoral output growth. Aggregate
investment largely exogenous.
(3) Foreign trade             endogenous                    partly endogenous       Korea: import coefficients. Exporters; profit
maximizing subject to targets.
BACHUE: imports endogenous.
(4) Sectoral disaggregation   29 seclors                    13 sectors              Korea: also 4 firm (or farm) sizes in each sector.
BACHUE: also distinguishes 4 types: modern,
traditional, rural, and urban.
(5) Production functions      neoclassical                  not included            Korea: uses two-level CES or Cobb-Douglas
Functions for labor and capital.
BACH U E: sectoral output demand determined
subject to aggregate constraint. Both use 1-0
coefficients for intermediate goods.
(61 Technical change          exogenous                     exogenous               Korea: neutral and disembodied in production
functions.
BACH 'iE: exogenous change in factor
productivity.
,7) Aggregate output          endogenous                    usuall) exogenous       Korea: aggregated from solution of model.
BACHUE: aggregate growth rate usually
imposed exogenously, but can be demand
driven or subject to other constraints.



(8) Inflation                endogenous                   not included           Korea: inflation determined by model of
financial systems and monetar; effects.
(9) Labor supply             usually exogenous            endogenous             Korea: can use aggregate labor supply
functions by skill category.
BACHUE: disaggre qated behavioral functions
derived from household decision model.
(10) Labor mobilitc          endogenous                   endogenous             Korea: by skill categories among sectors,
achieving labor market equilibrium.
BACHUE: adaptive over time as a behavioral
function of wage and income dilTerentials.
(HI) Unemployment            usually exogenous            endogenous             BACHUE: function of labor market
expectations and outcomes.
(12) Self emplo ed           treated specially            treated speciallx      Korea: immobile within periods. Service sectors
partially absorb or release labor.
BACHUE: limited mobility in modern sectors,
absorptive in traditional sectors.
(12) Wage employment         endogenous                   endogenou!             Korea: from solution of CGE model.
BACHUE: from labor allocation, labor
demand, and mobility functions.
(14) Wages and prices        both endogenous              wages and agricultural  Korea: determined by supply-demand
terms of trade endogenous  interactions in CGE model.
BACHUE: wages determined by value added
shares and labor utilizations with interperiod
quantity adjustments. Terms of trade
depend on excess demand for food.
(15) Worker composition of   endogenous                   endogenous             Similar technique used in both models.
households
(continued)



Table 3: Continued.
Features of the Model                Korea                     BACHUE                            Comments
(16) Income distribution     endogenous                   endogenuus             Korea: overall distribution (by households and
workers) derived from size distributions
within 15 income recipient categories.
BACHUE: overall distributions derived from
size distributions within 13 production
sectors and 4 income recipient categories.
(17) Remittances and transfers  endogenous                endogenous             Korea: fixed ratios.
BACHUE: functions of migration and income
differentials.
(18) MTigration              endogenous                   endogenous             Korea: function of rural-urban income
differentials (with upper bounds).
BACHUE: derived from household decision
model using micro functions of migrant
characteristics, macro functions of rural-
urban wage differenials and regional income
distributions.
(19) Fertility               not included                 endogenous             Korea: net labor force growth and skill
compositing are exogenous.
BACHUE: part of household decision function.
(20) Mortality               not includedl                endogenous             BACHUE: function of income and income
distribution.
(21) Education               exogenous                    endogenous             Korea: exogenous trends in skill composition of
labor force.
BACHUE: submodel of educational system.
(22) Population size and     exogenous                    endogenous             Korea: net population growtlh spdcified
structure                                                                       exogenously. No age, sex composition.
BACHUE: population tracked by age, sex,
region, and education.



(23) Optimization           by consumers                not explicit           Korea: profit maximizing producers, utility
by producers                                          maximizing consumers. Both price
responsive.
BACHUE household model implicitly based on
optimizing behavior.
(24) Solution               simultaneous/recursive      recursive              Korea: behavioral within-period closure
through price and wage adjustments
maintains consistency among firm,
household, government, and trade accounts.
Linked to inter-period dynamic model.
BACHUE: solution almost entirely recursive.
Consistency attained by wage and quantity
adjustments linked to dynamic behavioral
response.



78                                                1. Adelman et al.
are both economny-wide models with Lluite detailed speciflications of the
flow of income to recipient groups. Thev are botth dynamlic and are
recursive in time. They both can be seen as extensions of dvnamic multi-
sector planning models in that they include more realistic behavioral
speilfications and they inco.rporate as e-w:i.)genous variables m.any more
of the components of Stone's social accounting svstem.
While similar in broad approach. the two models are however \erv
different in their detailed structures and specifications (as Table 3
indicates). While some of the differences arise from the fact that the
models are for different countries, most arise from their different time
frames and extent of demographic concerns. Thus, while BACHUE
includes economic feedbacks on denmo,graphiic phenomena. the Korea
model excludes them. On the other hand, while the Korea model
includes in its economic sub-sector demand and si- p,plv interactions and
their feedbacks on   produCtion. resource allocation, and  prices,
BACHUE is purely deemand driven, with exogenous growth targets for
aggregate output. Both models share an input-output description of
intermediate dem-nands. However, the Korea mnodel hals neoclassical
production functions while BACHtUE uses linear cetelicteient. Tlihe Korea
model is, in general more neoclassical and market-oriented than is
BACHUE. In terms of dynamic behavior, BACHUE is a pure disequi-
librium growth model while the Korea model mixes within-perio d market
clearing equilibrium subject t) initial conditions with a dis,eqIuilibLium.,
quantity adjustment, d3 nanmic model. It would appear that while there
are similarities between the two model>. their basic airchitectulre and
behavioral specifications are sufficientl\ different so that one would not
conclude that any similarity in results came aboout purely because of
similarities in model structure.
The earlier discussion of policy environ nientes stressed the similarities
and differences between the two economies. Both Coun11tries are of
moderate size, are open, have hihlNh literate p0pu1lahtions. and--in recent
vears--have pursued similar development strate,ies. In contrast. the
institutional structure in the Il iculinral sectors is quite ditlTrenr in the
two countries. The industrial base is muclh smaller in the Philippines, less
diversified, and the economy has not completed the first phase of import
substitution. TIhe cultural traditions are di flerent. Korea ha11. un1der,one
a denmographic transition while the Philippinies has not. Thus, there zare
substantial differences between the two economiies. If the siniilaritk of
experimental results arises from sinmilarities betwecn the two coulntries, it
seems unlikely that it is due l)Urelk to similalrity in size or degree of
openniess. Similar-ity in past emiphal,is onl edLucation and in recenit
development strategy might well pl\ sonie role in the similarity of



INCOME DISTRIBLTTION PLANNING MODELS                                79
results, at least for some of the ex.perinmen s. However, many of the more
general results concerning the importance of certain key variables sLuchI
as migration and the aoriCUltUral terms of trade clearlv do not depend
on development strategy or initial conditions.
Finally, one can also discard the hyipothiesis of puire coinicidenice. Thle
model responses are too similar on too many poinlts and in too many
experiments to support the conitentioni of acciden tal congruence. One is
therefore left with the hvpothesis that the models capture something
"real" about fundamental interactions in a broad class of less developed
countries.
5. CONCLUSION
Both models suggest rather strongly that:
(1) The size distribution of income is e.v (eedinZgZv stable, even in the face
of substantial policy interventions. Whether or not the intervention is
sustained over time, it is rare that there is more than a 5'( change in the
Gini coefficient, even in the long term. or thiat a percentile's share is
altered by more than 15`%. Most single-policy interventions. even when
quite large, do not have significanit lasting effects. Onl wlvheni a stufficient
number of different interven tions are applried simultaneously, so that
there is, in effect, a change in developmilent strategy. are mole sizeable or
lasting effects possible. These results support the position that structural
change is required to affect ineLualit% and that equLlit\ objectives must
shape the choice of basic developmient strategy if they are to be met.
(2) Thle relative positioni of la/iou  soc io-cIn tic goi i.s s more sensitiVe
to the chloice of econofloic polity theanl is the overall dli.vribllioll. By
appropriate choice of instruments, it is possible to favor or discriminate
against particular groups. The rehlti'e degrees of poverty and wealth in
the economv as a whole are usually little affected. but the comlrosition
of the poor and wealthy groupLN can c1hanLgCe substal]n111t111. This applies
particularlv to the urban-rural balance of poverty.
(3) The agricultural terms of I/cielt' clLe the most i/lmporta/ii single targ-et
fior polic n inervenntio to improve incomne eli%trihtion. because of their role
in the determination of agricultural hous,ehold inconmes, anid because of
their sensitivity to a vairictN of appa rentlv Uinrela ted policy chatnuges. An
increase in the terms of trade rmiaes the incomnes of small farmllers and, to
a lesser extent, landless labor, and injures urban -roupss inclUdilllg the
urban poor. Overall the net efrect is to reduce po\erty even thoUgh'l the
incomes of rich farnmers are raised.
(4) lVi raiuna is the mtfoS't imlp)Urt i)t (h'1J1))IZrapl/hic ltiriabhh /for iflCOlcm  and
employvment polici up to (t least a 25 re('cr lii'iLroi. It can affect poverty
significantly. In general, the incOlmes of the urban poor are muchi highler



80                                                    1. Adelman et al.
than the incomes of the rural poor. so that policies to increase rural to
urban migration by making migration easier alleviate overall poverty.
Out-migration also improves the agricultural terms of trade, further
contributing to income equalisation and improvements in rural incomes.
(5) Population growth limitation has relativelv little effect on poverty,
reduction over a 25 vear hotizon. Indieed, there is a slight tendency for
population policies to lead to a deterioratinig income distribution and
increasing poverty in the short to medium term. This is because reduced
family size reduces demand for food, and thus tends to lead to lower
relative food prices. Eventuall . the terms of trade turn towards agricul-
ture, as rural labor supply is restricted. Beneficial income eflects then
emerge. But this occurs only in the very long term. The effects on
population growth of egalitarian development strategies. by contrast, are
likely to be significant-perhaps reducing fertility more than convention-
al family planning programs.
(6) Appropriate trade strategy cani lead to signiticant improvements in the
share and absolutie incomcs of the poor, given appropriate initial condi-
tions. The difference to the poor between following a labor-intensive
export strategy and an import-substitLtionl strategy can be major and
lasting. The major impact of trade strategy is not its direct effect upon
urban wages and emplovment. Rathler, it is the leakages to rural groups,
through induced chanLes in micration and terms of trade effects, that
provide the major influence of trade strategy upon poverty and equity.
A labor-intensive export promotion package leads to an improvement in
the agricultural terms of trade, thereby improving the distribution of
income and reducing poverty througgh trickle-down to the poor farmers.
An import-substitution strategy is likely to have the opposite effects.
(7) .afonv other policies have IInL'.vpeCIL'd etfects, or have exvpected effects
for une.xpected reasonis. Rural works programns tend to reduce rural
poverty, but may also reduce urban p1overt%, and indirect effects-i.e.,
those felt elswhere than in rural conistruction --are more important than
direct. Technology cholice does not on its own significantly affect income
distribution, appropriate technologx is difficult to design, even in princi-
ple. Small scale industry policies have only limited effects because
changes of an adequate size are very dificult to make. Transfers are
potentially effective as a means of redittributillg incoi-ne-but only
continued transfers of a magnitude quite inmplausiblc unider existing
political and institutional constraints. An overall rural development
strategN cani be moderatelv successful in raking, rural incomes but after
self-sufficiency in food is attained. this reqLuires agriculturall productivity
growth to be limited, or measures taken to increase food demand
rapidly, or price support policies to be inistituted.



INCOME DISTRIBUTION PLANNING MODELS                           81
(8) Balanced, across the board policies are nieeded for effective anti-
poverty policy even more than for economic grow,th. A big-push, balanced
strategy which impinges on many sectors and target groups is best. The
distributional analogue to the "growth pole" strategy is either ineffective
or tends to dissipate rather quickly. Experiments with extreme balanced
strategies in both models produced results which suggested that moder-
ate movements towards equality were feasible, but that these would
require major changes in development policy, and might well require
structural changes in social and economic relationships.
(9) Partial analyses often lead to incorrect conclusions on the ultimate
effects of policy interventions. Once the impact of any intervention has
permeated throughout the system. there can be differences not only in
orders of magnitude, but also in directions of effect. Moreover, interac-
tion effects among policies are important, so that the outcomes of
combinations of policies are rarely equal to the sum of their components.
Thus policy experiments with packages and global strategies show that
well-designed overall programs are liable to be considerably more
effective than analysis of each of their constituent policy changes would
suggest. These conclusions demonstrate the need for careful, system-
wide analysis of the impact of potential policy actions-analysis which
takes account of indirect and dynamic effects of policy intervention.
There are evident qualifications to make to the conclusions we have
reached. They derive from models which are specific to the socio-
economic structures of the countries concerned. Moreover, the outcomes
for the Philippines and South Korea rest on the existence of certain
preconditions in terms of the ownership of land and the distribution of
human and physical capital. These are prior to the experiments dis-
cussed, and in other countries where these preconditions are not met
appropriate structural changes would be required before comparable
policy results can be expected. Finally, as is the case for all quantitative
models and despite their enormous size, they contain only a fraction of
the relationships which we think are relevant, either because necessary
data do not exist or because our knowledge is inadequate. Nevertheless,
it is clear that the benefit in terms of insight vastly outweighs the cost in
terms of complexity. And the similarity in results from the two models
lend significant reinforcement to conclusions drawn on the basis of each
model taken separately.
The general conclusion for income distribution policy must focus on
the size of the effort required to achieve sig-nificant results. Only a
massive, wide-ranging, balanced and continued attack on poverty and
maldistribution of income has much chance of succeeding: lesser
modifications to existing strategies will fail. On the other hand, our



82                                                          I. Adelman et al.
analysis indicates that strategies for successful planning of income
distribution can be devised using an integrated array of policies, without
changing the fundamental rules of operation of the economy. The goal
is therefore not impossible; the policy instruments exist. If they are not
used it is because of the power of privilege and of wealth. More
fundamental changes in social structure and institutions are likely to be
needed, not necessarily for their own sake, but because without them the
power to implement the necessary economic policies will be lacking.
REFERENCES
Adelman, I., and Robinson, S. (1978a) Income Distribution Policy in Developing Countries:
A Case Study of Korea. Stanford: Stanford University Press.
Adelman, I., and Robinson, S. (1978b) Income Distribution, Import Substitution, and
Growth Strategies in a Developing Country, in Adaptive Models (R. Day, ed.).
Amsterdam: North Holland.
Adelman, I., and Robinson, S. (1977) Migration, Demographic Change and Income
Distribution in a Model of a Developing Country, Princeton University, Research
Program in Development Studies, Discussion Paper No. 71.
Blandy R. (1974) The Welfare Analysis of Fertility Reduction, Economic Journal (March
1974).
Brown, A., and Deaton, A. (1972) Models of Consumer Behavior: A Survey, Economic
Journal, Vol. 82.
Hopkins, M. J. D., Rodgers. G. B., and Wery, R. (1976) Evaluating a Basic Needs Strategy
and Population Policies, International Labor Labor Review,
International Labor Organization (1974) Sharing in Development in the Philippines. Geneva:
I.L.O,
Pyatt, G., and Roe, A. R. (1978) Social Accounting Methods for Development Policy and
Planning: A Case Study of Sri Lanka. London: Cambridge University Press.
Pyatt, G., and Thorbecke, E. (1976) Planning Techniques for a Better Fiuture. Geneva: I.L.O.
Robinson S. (1976a) Toward an Adequate Long-Run Model of Income Distribution and
Economic Development. American Econonmic Review 66, No. 2.
Robinson S. (1976b) Income Distribution Within Groups, Among Groups, and Overall: A
Technique of Analysis, Princeton L;niv*ersiry. Research Program in Development
Studies, Discussion Paper No. 65.
Rodgers, G. B., Hopkins, St. J. D., and Wery, R. (1977) Economic-Demographic Modeling
for Development Planning: BACHUE-Philippines. Geneva: I.L.O.
Rodgers, G. B., and Wery, R. (1977) The Effects of Economic Policy on Fertility,
Population and Employment Working Paper No. 53. Geneva: I.L.O.
Schultz T. P. (1974) Fertilitv Determines a Theory, Evidence and an Application to Policy
Evaluation. Santa Monica, CA: Rand.
United Nations (1973) The Determinants and Consequences Of Populat ion Trends, Vol. 1.
New York: U.N.
United Nations (1975) Towards a Srsietn of Social and Demographic Statistics. New York.