                                          PRELIMINARY
                                          VERSION, 12 June 1991
 
 
                A MACRO MODEL FOR A SMALL ECONOMY
 
 
SECOND PART  (See file Suriname1.txt for first part)
__________________________________________________________________
 
                    CONSTRUCTION OF THE MODEL
 
 
Micro price equations
 
For each of  the eleven  products in the  micro-block, an equation
was devised for  the years from  1954 up to and  including 1987 in
which the market price was derived  from the world market price of
the product  concerned.   The  comparisons  are first  set  out in
absolute levels and  then changes are  shown via the cointegration
approach (Ten  Cate  and  Draper,  1989).   In  this  way,  it was
explicitly  taken  into  account  that  Suriname's  exporters  are
price-takers on the  world market.   Where Suriname's export price
for a particular year deviated  from the world market price, there
was usually a catching-up effect the following year.
 
Micro production equations and investment equations
 
For each  product, the  price cost  ratios were  first calculated.
Basically, it boils down to the numerator being the selling price,
and the denominator  containing a  weighted average  of wage costs
per unit of output  and the import costs  per unit of output. (The
formalization  is  possible   because  intermediate  supplies  and
capital goods ultimately  also consists  of wage  costs and import
costs, plus  an  increase  to  take  account  of  interest).   The
proportion  of  wages  to  import  costs  varies  from  product to
product.
 
For each of  the eleven  products in the  micro-block, an equation
was devised in which the level  of output (or change therein) is a
function of the price/cost ratio.   This function is different for
each product.
 
For agricultural  crops,  an  intermediate  stage  was  added: the
planned acreage was  stated from  the average price,  and then the
exports  (after   subtraction   of   domestic   consumption)  were
calculated from the level of the planned acreage.
 
Furthermore, per product (and in a different way for each product)
the level of investment was derived  from the change in the change
in the level of output (see  Chapter 5 for extensive discussion of
this point).
 
The overall micro-block
 
After addition of  all the  results of the  micro price equations,
the average export price is derived.
 
On the  basis  of  the  results  of  the  eleven  micro production
equations, the total export value was calculated.
 
Addition of the investment  in the micro block  gives the total of
investment in the export sector.
 
In the context of  this study, priority is  given to the modelling
of linkages.  Refining of  the production and investment equations
for each of the products in  the micro-block should be possible on
the basis  of the  specialized knowledge  of the  product experts.
The micro-block  should be  able to  be extended  on the  basis of
specialized knowledge to products which could be important for the
future, such as  palm oil,  tiles, porcelain  clay, natural stone,
mineral oil and gold.
 
Although  the  micro-block  could  certainly  be  improved  at the
component level, the block as a  whole appears to deliver a result
which is head and shoulders above  the result of an calculation of
the export price,  export value  and investments made  just at the
macro level.  This is clear from  chapter 5.  In that chapter, the
equations are made at  the macro level.   In doing that, it became
clear  that  the  three  equations  would  have  led  to incorrect
results.
This is suggested by the estimate provided by a macro export value
equation, i.e. that  the change in  the competitive position would
not be relevant.  This is wrong, as our micro analysis shows.
 
Moreover, our  macro  calculation  of  the  export  price  gives a
coefficient of  only 0.58  for the  influence of  the world market
price.  The estimates at the  micro level, in which the problem of
heterogeneity plays no role, shows that it should be around 1, and
the significance of the result  at the micro level is considerably
higher.
 
A  pure  macro  estimate  of  investment  provided  no significant
explanatory variables.    A micro  approach  appears to  provide a
result, however.  In  that case, the  investment equation is first
calculated.  This  appears, together with  income from profits, to
lead to significant coefficients.
 
The conclusion is  that a pure  macro approach is too  crude to be
able  to  model  the  role  of  relative  prices  properly.    The
micro-block is therefore  essential.   A simple  macro model would
have given incorrect analysis and wrong simulation results.
 
This discovery resembles  the "discovery" in  the early 'seventies
that in Southern Suriname,  there was an  Indian tribe which still
appeared to be living in the Stone  Age.  For people living in the
big city,  this  was  something  entirely new.    For  the Indians
themselves, there was no question of "discovery", because they had
known about their own  existence all along.   In the same way, the
micro-economic behaviour which we recorded using the model was, of
course, well known  to the firms  concerned all the time.   From a
macro-economic viewpoint it  certainly seems to  be something new.
The incorporation  of a  micro-block  into a  macro model  has the
advantage that one  can look  at behaviour  by firms  from a macro
viewpoint, and that  one can also  examine the interaction between
micro and macro.
 
Macro-block
 
We shall only touch briefly here  on the macro-block of the model.
The  specification   of   most  behaviourial   equations   in  the
macro-block is mainly well  known, particularly inspired by, among
others, the  FREIA-KOMPAS model  (Van  den Berg  e.a. 1988).   The
literature cited in that  model (plus the  literature cited in the
literature etc.) gives  access to the  accumulated knowledge about
the construction of  macro-economic models.   Special attention is
required for  monetary  equations,  the  micro  foundation  of the
consumption equation, and the wage equation.
 
Money supply equation
 
The money supply equation is inspired  by the work of Caram (1981)
which concluded that the  behaviour of banking  in the creation of
money was influenced  by government.   The banks  in Suriname have
the function of  passing on  savers' funds  to firms (economically
neutral), and  endeavour  to make  a  contribution to  obtaining a
monetary equilibrium, which  can be  achieved where  the change in
the money  supply  is in  step  with the  development  of national
income.
 
Furthermore,  the  very  open  nature  of  Suriname's  economy  is
important, because an ex ante additional money supply leaks abroad
(at least, until 1984 - we shall return to this point).
 
By definition, change in  the money supply is  equal to the amount
of domestic monetary  financing by the  government, the balance of
payments (excluding  inter-bank  foreign  currency  transfers) and
increases in the  money supply  generated by banks.   The monetary
financing by  the  government  and  the  balance  of  payments are
endogenous in our model.  These  two sources of money creation are
explained in other  equations.  Therefore  only the money creation
by banking  remains to  be  explained.   From  the results  of the
calculation, it seems that at least  half of what money is created
by government  and  foreigners  deviating from  the  trend (growth
keeping pace with national income) is  retained by the banks.  The
very stable  relationship  between money  supply  and GDP  is only
partially explained by  the bankers' accommodating  attitude.  The
open nature  of  the economy  is  more  important.   As  a result,
monetary  financing  by  the  government  leads  to  money leaking
abroad.
 
Rate of exchange
 
After the loss of a  great part of the  income from mining and the
suspension of  development  aid, the  government  only implemented
cuts to  a limited  extent.   Monetary financing  initially led to
foreign currency flowing abroad.  As the foreign currency reserves
were exhausted  in 1984,  the government  imposed an  import quota
policy.    Imports  from  1985  onward  would  appear  to be  well
described by the following semi-behaviourial equation: imports are
equal to  the  value  of  exports plus  the  balance  of  trade in
invisibles and the  balance of  the capital account.   This policy
does not limit  demand (income)  but rather the  supply (volume of
imports).  When demand exceeds  supply, prices rise.  Scarce goods
are sold  for a  higher price  than is  warranted by  the official
exchange rate.  Consumers always have more money than the value of
the goods which firms  are able to supply  to them.  Alongside the
official market, where Suriname guilders  had a fixed rate against
the dollar,  there  was  a  black market.    We  believe  that, in
principle, it is impossible to  forecast the exchange rate, but it
is  basically  easy  to  explain,  except  for  sudden  changes in
confidence in  Suriname's  currency.   The  black market  has only
existed for a  few years.   In practice,  there are  too few years
available for  observation to  produce a  time series calculation.
For the exchange  rate, we must  therefore make do  with a formula
which is based on  the idea that fluctuations  in the black market
exchange rate  are  a function  of  the changes  in  the liquidity
ratio, and the level of that  ratio as a deviation from its normal
level.
 
In addition  to this,  all sorts  of confidence  factors come into
play, but these cannot be properly quantified.
 
Consumption function
 
Calculation of the  macro-consumption equation for  the years 1955
to 1984 inclusive did indeed provide a significant coefficient for
earned income amounting to 0.96. For profits, there is no data set
available for  an analysis  at  a micro  level.   Further analysis
shows  that  there  are  no  significant  differences  between the
consumption functions  of the  various population  groups (Creole,
Hindu, Javanese and others).
 
Thanks to the micro foundation of the consumption function, it can
be  specified   unambiguously  in   advance.     Where  there  are
significant calculation results,  they can be  deemed to be useful
for forecasting, at any  rate, as long as  there are no structural
changes like the  introduction of import  quotas in 1985.    It is
not possible  to specify  all equations  using a  micro basis, and
where following  the calculation  there are  figures which  can be
deemed to be useful for  forecasting.  So in  the case of the wage
equation in Suriname, the specification is not fixed and therefore
the results of the calculation must be considered less meaningful.
 
Wage equation
 
 
Although  the  consumption  function  per  individual  can deviate
considerably from  one year  to  another, the  average consumption
ratio from  earned  income  (and  from  income  from  profits too)
nevertheless remains stable thanks to the law of large numbers.
This law does  not apply  to wage rate  determination in Suriname.
The wage  equation in  the case  of Suriname  cannot have  a micro
foundation.    Wage  rates  are  not  determined  by  thousands of
individuals acting independently, but are defined institutionally.
Unlike the consumption equation, wage equations are not considered
useful  for  forecasting,  but  are  nevertheless  suitable  as  a
semi-behaviourial  equation,  a  tool  for  the  simulation  of  a
reference path.
 
It is true  that there  are thousands of  employers and employees,
but  the  modification   of  wage   levels  is   dominated  by  an
institutional process at  a national  level.   The social partners
who have to  give their  agreement are  not only  motivated by the
change in  retail prices  and  unemployment.   A number  of social
considerations and  political influences  play an  important role,
while inflation and unemployment do not exercise a simple, stable,
linear influence.   So, for example,  in 1986 and  1987, the trade
unions accepted  nominally  virtually unchanged  wages,  while the
average (including black  market) consumer  prices doubled between
1985 and 1987.
  Especially with  wage  equations,  one should  realize  that the
science of  economics is  a social  science.   The objects  of our
research are different from the  objects in natural sciences.  The
objects studied in  economics are  subjects at the  same time, and
can  do  something  very  special:  talk  about  their  (intended)
behaviour.  So, social partners are able to adjust their behaviour
in the light of  the results of the  first simulation.  Because of
that aspect of wage determination, an economic model should not be
used mechanically.  Provisional results  from the model are rather
important as a  subject of discussion.   The results  of the first
simulation  can  be  submitted  to  social  partners.    The first
simulation  can  be   made  with   the  wage   equation  (and  the
behaviourial equations concerning  the government,  which we shall
discuss later) with  coefficients which  reflect the  result of an
calculation of  the  past.   This  is a  forecast which  is  not a
prognosis.  It can even happen that social partners and government
may decide, after studying the  results of the first simulation to
modify their behaviour, with the  aim of changing the developments
which are predicted  in the first  simulation.  Communication with
social partners and government  in Suriname over  the results of a
first  preliminary  calculation  over   the  years  1985  to  2001
inclusive fell outside  the framework  of this  study.   We do not
refer to this  preliminary calculation as  a prognosis, but rather
as a  reference  path, because  it  has great  significance  as an
illustration of the  operation of  the economy  in accordance with
our model, but no value as a prognosis.
 
Institutional (semi-behaviourial) equations for government actions
 
In our  study,  it  appeared that  the  most  important government
variables in  the past  had very  stable relationships  with other
economic variables.   This is  true for various  revenues, such as
levies on bauxite,  import duties, excise  duties etc.  Government
expenditure  also  appears  to  have  a  relationship  with  other
variables.  So, government  investment had a  very close link with
development aid,  the interest  paid is,  as might  be expected, a
function of prevailing interest rates and government debt, and the
latter increases as a  result of the  finance deficit.  Concerning
the explanation of the  size of the  deficit, a distinction should
be made between the years before and after 1982.  Before 1982, net
material government consumption could be explained on the basis of
the difference between revenue and wages, as a result of which the
finance deficit is zero, apart  from temporary fluctuations.  From
1982 onward, government  consumption is exogenous,  and is totally
unrelated to revenues,  and the finance  deficit is the remainder,
which deviates  sharply from  zero.   Apart from  this fundamental
change in behaviour, it  appears that in the  past, there was such
stability in the  behaviour of government  that we ultimately used
the version of our  model in which  government variables were used
via institutional (semi-behaviourial)  equations, and therefore as
endogenous.   The  historical  simulations were  also  carried out
using that model.
 
Overview of the model
 
We endeavoured to develop the  most concise model possible, and we
succeeded in  restricting the  number  of variables  to 246.   The
MACMIC model therefore has 246 endogenous variables, comprising 57
estimated behaviourial  variables  and  189  definition variables.
Furthermore, the number  of exogenous variables  can be reduced to
49, including  a number  of dummies  and a  few variables  such as
depreciation, which  play  no  role  in  the  model  and  are only
included to  provide  a smooth  link  with the  accounting program
MACROABC,  which  produced  the  National  Accounts  and  Monetary
Statistics.   The key  exogenous variables  are: the  world market
prices for  Suriname's  export products,  the  level of  prices of
imports,   the   potential   active   population   and  migration,
development aid, and a few  exogenous government variables.   Thus
the total number of variables in the strictest sense has been kept
below 300.  By  doing this, not only  is the requirement fulfilled
for  ease  of  comparability  which  is  especially  necessary  in
developing  countries,  according  to  the  one-time  pioneer, now
eminence grise in the field,  J.J. Polak (see Driehuis a.o. 1988),
but a  practical  purpose is  also  served.   With  less  than 300
variables, it is quite feasible to  run the model using a software
package  which  according  to  Van  Nes  and  Ten  Cate (1989)  is
convenient and easy to learn, and  which runs on a simple personal
computer.
 
The core of  the model  consists of behaviourial  equations in the
micro-block and the  macro-behaviourial equations  (see diagram at
the end  of  Chapter  5).   In  addition  to this,  there  are the
institutional (semi-behaviourial) equations for the government and
many definition equations.
 
The model is estimated  on the basis of  data from the micro macro
data set for  the years  1954 up  to and  including 1987.   In the
model, there  are  both  dummies and  autonomous  terms.   Dummies
concern incidents which we  cannot predict, but  which are for the
explanation  of  the   variables.    So,   for  example,  guerilla
activities can explain a  sudden drop in  bauxite production.  For
this  reason,  before  moving  on  to  the  final  version of  the
calculations, dummies were  added to various  equations.  Although
there  are  many  incidents  in  Suriname,  partly  for  political
reasons, partly  because of  the  small size  of the  country, the
number of dummies can be restricted.   So, only five of the eleven
behaviourial equations in the micro-block have a dummy.  After the
calculation  of  the  equations,  or  after  the  first  round  of
historical  simulations,  residues  seemed  to  arise,  for  which
explanations could still  be found.   Only if  the explanation was
plausible  and  some  idea  could  be  obtained  of  the order  of
magnitude (independent of the size  of the residue, of course) was
an autonomous term produced.  These autonomous terms were not used
to re-calculate the estimates from the equations.  All the results
of the calculations  and the  associated graphics  in this studies
are exclusively autonomous  terms.   The purpose of  this study is
not to  achieve  the  narrowest  possible  difference  between the
results of the  calculations and  the actual figures,  but to gain
insight into linkages, and part of this involves gaining a clearer
view of what  we do not  understand properly.   There is certainly
material  for  further  study.    The  autonomous  terms  are only
included in the historical simulations  and per equation per year.
The matter  of  whether  an  autonomous  term  has  been  added is
meticulously documented.   In this  way, it was  possible to avoid
that a  plausible  deviation in  a  specific equation  in  a given
simulated year should distort  the entire picture.   In the model,
all the equations are  closely inter-related, and  an error in one
year can have repercussions throughout  the model even in that one
year, and  subsequently  the  errors  in  all  the  variables work
through into the following years in a cumulative way.
 
 
Instructions for use
 
The restriction of  the number  of variables  in the  model in the
strictest sense to just 295 is not problematic, because the output
can be subjected to further  processing using an easily accessible
spreadsheet program.  Using that  program, many more variables can
be distinguished effortlessly.   Refer to  the MACROSA spreadsheet
program mentioned earlier, in the  micro macro data set, with data
provided for the period 1954-87,  with the option of completing it
with data for 1988 to 2001, using the output from MACMIC.
 
The  model   consists  of   a   micro-block,  a   macro-block  and
semi-behavioural equation block, plus a number of auxiliary blocks
and  batch  files.    Refer  to  the  micro  macro  data  set  for
instructions, and order  a diskette with  the demonstration of our
model.   Using  this,  simulations  can be  carried  out  at will.
Actually using  the  result of  this  study will  enable  far more
detailed knowledge to be obtained than in possible just by reading
this text.
 
 
                            SIMULATION
 
 
Historical simulations
 
The period  1954-87  is  split  mainly on  the  basis  of economic
turning points into  five component  periods, each  of about seven
years, i.e.:  1954-1960; 1961-67;  1968-75;  1976-82; 1983-87.   A
historic simulation was made for each of these five periods.
 
For example: for  the period  1954-60 we  started with information
concerning the year 1954, but no information for later years, with
the exception of the exogenous variables.  So only a few variables
were used  for the  period 1955-60,  such as  world market prices,
import prices, development  aid and population  growth.  Using the
MACMIC model, an  estimate was then  made for the year  1955.  The
output from that then provided the  input for the estimate for the
year 1956, and so on until 1960.
 
This battery of historical simulations was carried out for each of
the five  component  periods.    These  are  severe  tests indeed,
because errors  are  cumulative (an  error  in the  equation works
through to the next equation,  etc. etc.).  In addition, exogenous
government variables  are  used.    For  an  overview  of  the key
variables with the dicrepancies between the actual figures and the
results of  the  historic simulations,  we  refer to  the graphics
(graphic  6.3.1.  onward).    These  compare  absolute  levels and
changes. The inequality  coefficients of twelve  key variables are
calculated (see  table  6.3.2.).   These  are  percentage changes.
Overall, the inequality  coefficient appears  to be below  0.7.  A
less satisfactory  result was  returned for  the first  and second
period, for the consumption,  investment and employment functions.
The latter variable  was substantially overestimated  in 1971, and
substantially underestimated in 1975.  It is not beyond the realms
of possibility  that this  discrepancy  originates not  within the
model, but rather  due to  problems with the  exact measurement of
this indicator.
 
The average inequality  coefficient in each  period always appears
to increase  as one  moves towards  the  end of  the period.   The
usefulness for forecasting of the  model decreases in the same way
as a weather forecast, as one attempts to look further and further
ahead.  If one  uses the model for  simulation of policy variables
rather than  for prognosis,  the model  appears more  suitable for
medium-term simulations, because the model is calculated in such a
way that it takes  into account the  phenomenon that variations in
one year will be followed by variations in the following year.  It
is better  to consider  the  cumulative effect  over a  few years.
Moreover, its application  to medium-term analyses  is more useful
that for a single  year, because the  fundamental processes in the
economy (via price/cost  ratios) can take  several years to become
apparent.
 
These historical simulations do not  just constitue a test for the
model, but  are  also  the  basis  for  the  analysis  of mutually
dependent economic indicators per component period.  See the final
section of Chapter  6 from  which we  have summarized  some of the
conclusions:
 
In the  period 1954-60,  price/cost ratios  were favourable; firms
were investing  heavily; the  consumption ratio  of the government
was low; rice exports  tripled, while import-substituting industry
produced new products; employment in firms increased on average by
6% per year; this period could be classified as one of "growth".
 
In  the  period  1961-67  the  "Brokopondo  push"  occurred.   The
construction  of  the  storage   reservoir  and  the  alumina  and
aluminium plant.    Due  to  the very  open  nature  of Suriname's
economy,  and  the  foreign  financing  of  the  push,  there  was
initially no influence on inflation.  Increases in productivity in
the bauxite sector led  to wage rises  which spread throughout the
economy.  In agriculture, the government invested according to its
plans,  and  there  was   further  growth  in  import-substituting
industries.  Purchasing  power (a real  indicator) rose on average
by 9% per year.  This period with the push in the miniing industry
served as a  model for later  development plans.   No interest was
shown in factors related to business economics, such as price/cost
ratios, whereas  favourable price/cost  ratios  were the  basis of
growth in the 'fifties and 'sixties.
 
The period 1968-75  can be  classified as one  of "modest growth".
This  was  not  growth  driven  by  real  increases  in  corporate
investment, but rather growth  driven by increased consumption and
development  aid.     The   latter   exhibited  a   shift  towards
consumption.
 
The period 1976-82 can  be summarized as "money  no object!".  The
bauxite levy and development aid  were flooding in at a tremendous
rate.  The gigantic financial possibilities led to wage inflation.
Price/cost ratios became  unfavourable, and there  was no increase
of employment in firms.
 
1982-1987 could be  described as  the "crash".   The vast revenues
from the bauxite levy  and development aid fell  away.  Instead of
cuts  and   reorganization  of   government  finances,   by  which
price/cost ratios would  have been returned  to viable levels, the
government  adopted  a  policy  combining  monetary  financing and
import quotas.  This  led to the  creation of a  black market.  It
was only due to wage restraint by  the trade union (at a time of a
65% fall in purchasing power!) that hyperinflation was avoided.
 
Reference path 1988-2001
 
An important economic variable  concerns changes in the population
aged 15-65.  Apart  from migration, beyond  fifteen years old this
can be  estimated accurately,  due  to the  high stability  of the
mortality rate,  whereas fluctuations  in the  birth rate  take 15
years to  have an  effect on  the size  of this  group.   For this
reason,  and  also  to  carry  out  a  longer-term  exercise,  the
reference path  lasts for  about twice  the length  of the periods
used for the historical simulations  (which were about seven years
each).   The  reference  path  is  a  forecast  made  using MACMIC
starting in 1988 and ending in the year 2001.  We do not attribute
any value to this reference path as  a prognosis.  Its real use is
in providing a tool which enables comparison with simulations.
 
A prognosis is not possible in  this context.  In the first place,
there is  the  problem  that  we  cannot  be  fully  aware  of the
exogenous  factors  in  the  world  situation  (international  raw
materials prices,  import  prices  etc.).    This  is  a perpetual
problem for economic  forecasting, but  in this case  there is the
additional problem that  we do not  have any realistic assumptions
about future changes  in the  behaviour of the  government, and we
have to rely on  semi-behavioural equations.  The  same is true of
wage estimates.    In  addition,  at  present  a  black  market is
operating for foreign  exchange in Suriname.   It is  true that we
could make  a forecast  for  the black  market exchange  rate, but
because of the  unpredictability of  changes in  confidence in the
guilder, forecasts  of the  rate of  exchange have  no value  as a
prognosis.
 
Policy simulation
 
The MICMAC model can  be used for the  simulation of the effect of
divergent trends in  a variable  in the  model for  one or several
consecutive years.    For  example,  the  effect  of  a  change in
international raw materials  prices.  That  sort of partial policy
variant is simulated by examining each variant in each of the five
component  periods  from   the  past,  plus   the  reference  path
1985-2001.  Because  the  starting  situation  in  each  period is
different and the model is not linear, the same variant can have a
different effect in  different periods.   In order  to analyze the
operation of  the economy  of  Suriname as  it has  been modelled,
initially partial  variants  are  simulated.    After  the partial
variants, where only  one, or at  most a few of  the variables are
introduced, later  policy  packages  are simulated  with  a larger
number  of  variables  where  the  values  are  changed  from  the
reference paths.    (The  reference paths  covering  the component
periods in  the past  make use  of the  results of  the historical
simulations).
 
The first  partial  variant  concerns  a  higher  price  level for
Suriname's export products,  but no change  in import prices, i.e.
in all years of a period,  export price levels would be 10% higher
than  the  levels  in  the  reference  paths.    This variant  has
favourable consequences for  the economy.   The  consequences of a
partial variant  with a  10% increase  in import  prices, but with
export price levels remaining the  same as in the reference paths,
are unfavourable.   The  combination of  these two  variants would
mean a 10% devaluation.  It  appears that the effects of these two
variants largely cancel each other out.  Here and there, there are
small plus and minus points, which differ greatly according to the
period under consideration.   However, there  is one special group
of variables - improving price/cost ratios.  In the periods before
1983, this led to employment  growth, but thereafter, there was no
effect.  The price/cost ratios became so unfavourable in the 1980s
that 10% devaluation is not enough  to bring them to a stage where
investment is viable  once again.   Devaluation certainly leads in
the years after  1985 to a  reduction in changes in  prices on the
black market,  but prices  increase  on the  official market.   If
wages react strongly  to the official  price increases, this leads
to an inflationary  spiral.   In the wage  equation proposed, this
does not happen because  it is assumed that  wages will only react
to a limited  extent to consumer  price increase, due  to the high
level of unemployment,  for which  reason the  weighted average of
the official and black market is proposed.
 
See Chapter  7  for a  number  of  other partial  variants.   With
reference to  that chapter,  we point  out that  wage restraint is
good for employment; an increase  in import duties has a different
effect on employment  in different  periods; government investment
in infrastructure  to support  rice cultivation  have a beneficial
effect in the  early periods  on growth and  employment, but after
1982 this is no longer the case, because the price/cost ratios are
so  low   that  firms   do   not  invest,   in  spite   of  better
infrastructure.  Finally, a  variant in which  the number of civil
servants at the  beginning of each  period falls by  3% instead of
growing which was what actually happened, and where the government
uses the salaries saved  for old age  pensions and child benefits,
has a beneficial effect  on employment prospects  in firms, but is
not enough to cause a fall in unemployment.
 
 
Half a century of simulation of development strategies
 
A development strategy is a  policy package with different partial
variants which is drawn up  by the government in consultation with
the  social  partners,   and  is  aimed   at  the  realization  of
development objectives.  As  this study is not  part of a planning
process, none  of the  development strategies  which are simulated
here have been borrowed from public institutions in Suriname.  The
strategies examined are  hypothetical in character,  and are taken
from  or  inspired  by  literature  in  the  field  of development
economics.
 
We examined the  recent discussions about  import substitution and
outward orientation (Lal, 1983;  Belassa, 1989; Koekoek, 1989; Kol
and Mennes, 1989;  Perkins, 1989; Robinson,  1989; Ruigrok, 1990),
Dutch  disease   (Van   Wijnbergen,  1984;   Van   Gemert,  1985),
application strategies  (Caram,  1985;  Khan,  1985);  Edwards and
Montiel, 1989; Landell-Mills,  1989; Meyer  and Vingerhoets, 1989)
and structuralism  (Taylor, 1983;  Jansen  and Vos,  1985; Janssen
1986; Stern, 1989).
 
For each period, a separate package  was simulated.  For the sixth
period (the  future), a  package  was formulated  on the  basis of
recommendations which  were  made  last  year  (mainly Weitenberg,
1989; Brahim, 1989; and World Bank, 1989).
 
In table  8.2,  the effect  of  each package  on a  number  of key
variables is summarized briefly.  These key variables usually show
percentage changes.   These  represent the  cumulative effect with
respect to the  reference path from  the starting year  to the end
year.  For  the last period,  a column was also  included with the
results of  the  simulation package,  i.e.  not compared  with the
reference path.   For  each package,  we summarize  below the main
points.   In the  first  package, applied  to the  period 1955-60,
government expenditure (in  the form of  a job creation programme)
and government  revenues  are increased  at  the same  time.   One
anticipates the possible occurrence of  the "Haavelmo effect".  In
earlier economic models where no supply factors were incorporated,
this should have led to economic growth.  In MACMIC, the influence
of prices and cost factors were not
 
Table 8.2. Six  packages over  six periods.   Effect compared with
the reference  path, cumulated  from  the   starting year,  in the
final year.
______________________________________________________________
                    '54- '61- '68- '76-'83-'88-          '88-
                    '60  '67  '75  '82 '87   '2001        '2001
                effect compared with reference path  total (a)
 
PRICES % change
Private consumption  -2    -2  -3   -11 -36  -33            381
Imports               0     0   0     0  21   -6            308
Wage base firms      19     1 -24    -6 -15    1            368
 
 
VOLUME % change
exports             -15     1   6    12  24   20            174
Private consumption   2     1  -3    13  99  -26             38
Investment by firms -12    14   6    35 385   -4           1360
Governm.consumption  -4    12  26    32  48  -42            -36
Imports              -9     4   7    10  28  -10            217
Employment in firms  -4    12   7     6  10   11             10
Purchasing power
     of wages        18     0  -5    18  50   87             27
 
VALUES % change
money supply        -20     0   2   -15 -52  -28            131
 
levels of
unemployment %        1    -2  -3    -2  -3   -7             31
wage ratio           13     0  -8    -2 -25   -9             35
price/cost ratio
  for rice        -0.12   0.0 0.1   0.5 1.4 -6.21
______________________________________________________________a)
the  last  column  does  not  concern  effect  compared  with  the
reference path, but rather the result of the simulation
 
 
overlooked, and  then this  package appears  to lead  to increased
unemployment after a few years.   Due to the detrimental effect on
price/cost ratios employment in firms  falls, in particular.  This
fall is even greater than  the increase in government jobs arising
from  the  job  creation.    (See  chapter  7  for  more  detailed
explanation of this and other packages).
 
The second package covers the period 1961-67.  It concerns raising
of  import  duties  on  goods  which  could  also  be produced  in
Suriname, and where there are  sufficient producers to ensure that
competition regulates the  market, and  eliminates the possibility
of  abnormal  profits  being  made.    Here,  one  could cite  the
importation of  potatoes  as  against the  domestic  production of
breadfruit, Chinese  tajer, napi,  pomtajer, and  bananas or rice,
products which, according to  research (ABS, 1983a) are considered
by the consumer as  a real alternative.   This domestic production
is still  expensive compared  to  imports, because  relatively low
tariffs were  applied.   This  package  provides for  a  freeze in
government jobs and  extra investment in  agriculture, financed by
the savings in salaries and a  limited increase in the taxation of
profits.   According  to  the  results  of  the  simulations, this
package  appears  to  have  a  beneficial  effect  on  growth  and
employment (see the second column with the figures in table 8.2).
 
In the third package, covering  the period 1968-75, we assume that
the substantial increase  in productivity in  the bauxite industry
is creamed off  through the imposition  of a bauxite  levy in 1968
(instead of  1974, as  actually  happened) and  that wages  in the
bauxite industry increase less than was actually the case, so that
the profits  from  bauxite  returned  the same  result  as  in the
reference path.  Pay restraint in the bauxite industry was assumed
to work  through  to  the rest  of  the  economy.   This  leads to
additional profits,  which  for  the  most  part  are  creamed off
through an increase  in the  taxation of profits.   The government
uses these  resources  partly  for  the  introduction  of  old age
pensions and  child  benefits in  1969  (actually, they  were only
introduced in 1974), partly for a reduction in import duties.  The
latter measure is beneficial for  price/cost ratios.  This package
is neutral for the financing  deficit.  The simulations indicate a
beneficial  effect  on  employment,  at  the  expense  of slightly
reduced purchasing power.
 
In the fourth period, which  covers 1976-82, there is a simulation
of what the effect  would have been if  a third of the development
aid had  not been  devoted to  government expenditure,  but rather
allocated to  pay  for  the  reduction  of  EC  import  duties for
Suriname's exports of rice, alumina, unprocessed wood and plywood,
and/or providing  subsidies  for  exports of  these  products from
Suriname.    Furthermore,  this  package  provides  that  a larger
portion of the remaining development aid should have been used for
additional investment in agricultural infrastructure, to encourage
investment by  individuals.   According  to the  simulations, this
package would have had a  beneficial effect on growth, employment,
and purchasing power.
 
In the package  for the  period 1983-87, we  considered what would
have happened if the government had not adopted a policy of import
quotas in 1984,  but had  instead reduced  its spending, increased
revenues and devalued the guilder  so that price/cost ratios could
have fallen to  a level where  investment became viable.   In this
package, it is assumed  that wages would be  frozen, which is only
possible when there  is full support  from the central government.
Implementation of this package would have led to considerably more
losses  in  purchasing  power  in  1984  and  1985  than  actually
occurred, but during 1986  the package would have  led to a slight
improvement, and in 1987 the  level of purchasing power enjoyed by
Suriname's working population and unearned income (such as old age
pensions) would have been  50% higher than  was actually the case.
Furthermore, this  package  would  have  resulted  initially  in a
slight  downturn  in  employment,  but   would  have  had  a  more
beneficial effect  from  1986  onward.    In  addition,  thanks to
healthier price/cost  ratios from  1986 onward,  a situation would
have  been  achieved  from  which  sustained  growth  and  falling
unemployment should have been possible.
 
For the period 1988-2001, a  package was formulated which could be
introduced in 1991.   The package  consists of various components.
To start with,  rules for  eliminating the  financial deficit, the
cause of the  present difficulties.   Government expenditure would
be  restrained,  and  revenues  increased  so  that  the financial
deficit would  be entirely  eliminated  from 1991.   In  this way,
revenues would  be  increased so  that  some expenditure  could be
increased instead of  decreased.  The  package provides for, among
others,  old  age  pensions,  child  benefit  and  benefits  being
quadrupled.  In addition,  rice prices would  be subsidized on the
domestic market, so that  rice prices to  the consumer in Suriname
remain the same, and prices to  the producer rise in proportion to
the percentage devaluation  - to be  discussed later.  Thereafter,
there are figures on how much the devaluation would have to be.
 
To do  this,  it is  first  necessary to  establish the  aim  of a
devaluation.     There   are   actually  two   aims,   namely  the
disappearance of the corrupting black market, and the re-alignment
of price/cost ratios at  a viable level.   The first requires that
there should  be  an increase  in  import prices  on  the official
market such  that,  with  the given  income,  demand  for imported
products does not exceed what can  be paid for by export earnings.
If we  assume  (we  shall  return  to  this)  that  wages  are not
increased, a  devaluation of  300% (a  quadrupling of  the rate of
exchange) would  be  necessary  to bring  price/cost  ratios  to a
favourable level.  Considering that the average exchange rate with
respect to the official rate in the reference path lies just below
1 to 4, such a devaluation  would be, in principle, well suited to
achieve an intermediate rate.  If a Lieftinck-style operation were
then to  be carried  out (we  mean a  temporary freeze  on savings
balances etc.) this  would avoid the  monetary overhang leading to
demand for imported goods.  Price  control is out of the question.
There is only  one sure way  of eliminating the  black market: the
Suriname  guilder  must  be   fully  convertible  against  foreign
currencies, and importation  must be  deregulated.   Even if there
are  no  currency  reserves,  it  is  possible  to  introduce full
convertibility,  provided   that  the   appropriate  measures  are
introduced  to  restrict  demand,  and  there  is  confidence that
monetary financing will disappear forever.
 
Confidence in the durability of the solution would be increased in
the Central Bank were to receive limited reserves in order to deal
with  discrepancies  which  can  occur  temporarily  through freak
circumstances.   Provisional calculations  in the  context of this
package suggest that an amount of around one hundred million Dutch
guilders per  year  over four  years  should be  sufficient.   The
Netherlands would appear to  be prepared to do  this, as part of a
coherent  package  (Hoekman,  1990).    Moreover,  such  action to
support the currency will only work if it is clear that it is only
foreign  support  to  a  policy  package  from  the  government of
Suriname.
 
We now return to the key  question in this package: is it feasible
for wages  to rise  in  accordance with  the reference  path under
these circumstances?    The  calculation -  i.e.  with  wage rises
assumed to be in line with the reference path - already results in
an increase in  purchasing power  of wages and  unearned income in
1991.
 
This is surprising  in the light  of the fall  in purchasing power
generated by  the policy  package  for the  period 1983-87  in its
first two years.   The explanation is that  people in Suriname had
already experienced a  major drop  in purchasing power  due to the
rise in  prices  on the  black  market.   In fact,  in  the policy
package for 1991, it is  just a matter of  prices on the black and
official markets equalling out.   In the case  of the former, they
fall, and in the latter, they rise.  Due to the multitude of price
movements,  both   upward  and   downward,   the  net   effect  of
implementing this  package is  very little  change in  the average
level of prices.  (The policy package for 1983-87 described above,
in contrast  returned  price increase  figures  for the  first two
years which were higher than  those which occurred in reality, but
after a few years, this policy package also leads to a lower level
of prices than actually occurred).
 
Using this  policy  package,  the prices  on  the  official market
became four times their previous  level, while prices on the black
market fell  by 50%.   Moreover,  people who  derived their income
from  the  black  market  saw  their  income  shrink dramatically.
Figures for  that income  do  not appear  in the  purchasing power
figures for workers  and recipients  of state  benefits.  Although
the average purchasing power of  those groups increases, it should
be borne  in mind  that those  workers who  were relatively better
able to buy  on the official  market (for example,  those who have
the time to  stand in the  queues) stood to lose,  while those who
had bought relatively more on the black market stood to make a net
gain in their  purchasing power.   We are  not able  to solve this
problem of redistribution here.   We would just  point out that we
are only testing a simulation package, and that the formulation of
a package  which has  a  satisfactory social  basis cannot  be the
objective of this study.  In assessing the social basis, it is not
only the  redistribution of  purchasing power  which is important,
but also  the effect  on  employment and  price/cost ratios.   The
simulation of this package generates a fall of 7 percentage points
in the rate of  unemployment.  Moreover,  the price/cost ratios of
existing  products  fall  to  such  a  favourable  level  that the
price/cost ratios for  new products -  not covered in  our model -
also become viable.   In that  case, perhaps more  growth could be
achieved than is shown by this simulation.
 
As  we  have  touched  upon   this  point,  the  question  of  the
limitations of the model should be borne constantly in mind.  This
macro micro  model is  primarily a  tool for  finding linkages and
making  a  coherent  analysis  of  economic  events  in  the past.
Furthermore, the model can serve as a tool to gain an idea of what
would  happen  in  the  future  if  certain  policy  measures were
adopted.   In  this  model,  the linkages  are  crucial,  but many
critical components have not been examined in depth.  Such a model
can only be  used in practice  as a complement and  a framework to
integrate the  results of  specialized studies.   Even  the export
sector which  was  modelled at  the  macro level  requires further
information based on the expertise of specialists on each product.
We therefore  warn against  considering the  devaluation figure of
300% (i.e.  75% less  foreign  currency for  1  Sf) as  a concrete
policy  proposal.    It  is  always  possible  that  more detailed
research on, for example, the price/cost ratios might show them to
be temporarily higher or lower than we were able to compute on the
basis of scant information.
 
If the devaluation percentage were to prove too low to achieve its
intended  purpose  (to  reach  viable  price/cost  ratios  and  an
official import price  which is  so high  that only  there is only
demand for that quantity  of imported goods which  can be paid for
by export  earnings), there  is  the risk  that  there will  be no
economic growth, the  black market  will continue  to exist, while
there is also the  risk that wages may  climb in response to price
increases on the official market, to reach approximately the level
of the percentage devaluation.  This would lead to an inflationary
spiral, and hyperinflation could be  anticipated.  It is certainly
not  true  that  things  could  not  get  worse  than the  present
situation.  On  the other hand,  there are also risks  that if the
percentage devaluation were too high.   In that case, prices would
rise more than is necessary, and this could be accompanied by wage
rises which  maintain  the  average purchasing  power  of workers.
Starting with supplementary wage rises is easy, but thereafter, it
is extremely difficult to control the wage-price spiral.
 
The restructuring  simulation using  MACMIC provides  two results.
It shows that a restructuring  programme can be implemented, leads
to  growth  and  would  even  maintain  average  purchasing power,
provided that it is supported by the social partners, particularly
in the  form  of  wage  restraint.    Another  conclusion  is that
attention must  be paid  to all  the inter-related  components and
that the calculation  of the correct  devaluation percentage needs
to be very  accurate.   This theoretical  simulation gives insight
into the working of the economy,  and especially leads to the view
that the model  can only be  used in practice  in conjunction with
thorough  micro-studies  and  in   consultation  with  the  social
partners.  Although in this  theoretical study, much attention has
been paid to the  empirical basis of the  model, we consider it to
be unsuitable  to  make  concrete policy  recommendations  from an
academic point of view.  There is one exception, however:
 
It  should  be  recommended  that  the  quality  of macro-economic
development strategies should  not be tested  out by experimenting
in real-life  economic situations.   The  people of  Suriname have
been live guinea-pigs for  this kind of  vivisection for too long.
It is better to test the ideas first, using a model.  We would add
at this point that our  initial restructuring simulations with the
test version  of the  model unfortunately  went wrong.   After one
re-specification, the  model  did  not  appear  to  be  leading to
convergence.  In another attempt,  the model crashed.  Ultimately,
these  laboratory  tests  showed  that  in  the  years  after  the
introduction of import  quotas, not only  the import equation, but
also the consumption  function were  quite different.   Only after
improvements to  the  model  based  on  this  idea,  and  the best
possible modelling of the black market  did the model appear to be
applicable  to  the  period  after  1985.    The  carrying out  of
"experiments" in  order  to  gain fuller  understanding  is better
carried  out  using  a  model,  because  then  the  carshes remain
confined to the model,  and a new variant  can always be developed
until the  subject  matter  is properly  understood.    Trying out
variants immediately in  the real-life situation  can also lead to
crashes, but they are more difficult to put right.
 
             'Een Macro-model van een Micro-economie'
 
                             ABSTRACT
 
This study  consists of  the construction  of an  empirical macro-
model of  a  micro-economy, the  small  and very  open  economy of
Suriname. The model thereof has been  used for the analysis of the
economic development of  Suriname and the  simulation of different
development strategies.
  If one tries to build  a model for a  small open economy, one is
confronted with the problem that the law of large numbers does not
work  in  the  export  sector.  Approximately  90%  of  exports is
furnished  by  only   eleven  products.  In   that  situation  the
development of  total export  prices, volumes,  and investments in
the export sector,  is dominated by  accidental factors. They make
it  difficult  to   get  the   underlying  economical  behavioural
relationships in the picture. However, we have solved this problem
by building a micro-block for the export sector, which consists of
special price, production and investment functions for everyone of
the eleven  main  export products.  Thanks  to the  method  of the
micro-block it  appeared  not  only  feasible  to  incorporate the
export sector of a small open economy in an empirical macro-model,
but it also  offers the opportunity  to build a  bridge across the
river, which divides  business-economics and macro-economics. With
this model one can study  the interaction between costs and prices
at the micro-level and the production, economic growth and incomes
on the macro-level.
  In the model the  "parallel market" is  also incorporated . This
market was  introduced  in  1984/1985 and  since  that  time other
import- and  consumption  functions  are  in  force  and  a double
exchange rate is operative.
  Further the model has  a government-block, with semi-behavioural
equations for incomes and expenditures of government.
  For the sake of  estimation of the equations  of the model, this
study started with  the construction of  a consistent dataset over
the  years  1954   up  to   1987.  See   the  special  publication
"Micromacrodataset" for a detailed description of this statistical
activity.  In  the  Micromacrodataset  are  also  reports  on  the
definition model "Macros(abc)", with  which the "Monetary Surveys"
and the "National  Accounts" are  generated. The Micromacrodataset
together with  Macros gives  all the  micro and  macro time series
which are needed to estimate the macro-model named "MACMIC".
  Most equations of MACMIC are estimated seperately, but some also
simultaneously.  Besides  that  we  tested  the  model  by  way of
historical simulations over  different periods:  54-60; 61-67; 68-
75; 76-82; 83-87.  The historical simulations  consist at the same
time of an analysis  of the economic  development and relations in
the Surinamese economy.  We also  made a  base line  for the years
1988 up to 2001.
  Based on  a  short survey  of  recent literature  on development
economics, different  policies are  formulated and  simulated with
the model.  Some  policy-packages are  also  simulated, especially
development strategies to  restructure the economy.  Thanks to the
micro-macro character  of the  model, the  role of  the prices and
parallel market can be taken into account explicitly.
  The preparation of concrete policy proposals is beyond the scope
of this theoretical  study. However,  the model and  data are also
available on  diskettes  (see Micromacrodataset).  This  gives the
opportunity to take cognizance  of the results of  this study in a
more intensive way, and to use the model for ones own analysis and
simulations.  The  dataset  is   available  for  work  on  further
improvement for the modelling of Suriname's economy.
  In the Micromacrodataset and  diskettes, which will be published
together  with   this  study,   one  can   find  a   basic  for  a
generalisation of  our data  handling software  and model.  It has
already been used for the phantasy island "Microland".
 
 
 
End of the   summary of "A   Macro Model of a   Micro Economy".
In MICLAND.TXT follows a  description  of how  our  Suriname model
can be adapted to other small, open economies: "Microland".
In MALTA.TXT  one can  read how  the  model has  been   adapted to
Malta.
 
 
