Chapter 3
The Planner and the Market:
The Takayama Judge Activity Model
The linear programming formulation of the Leontief input–output model, established
as the linear activity analysis model, represents an advancement in the construction of
applied general equilibrium models, because it introduces a great deal of flexibility
into the basic linear input–output structure. The lack of price-induced substitution
was overcome by the development of the linear activity model. By allowing inequal-
ity constraints and the introduction of an endogenous mechanism of choice among
alternative feasible solutions, the effects of sector capacity constraints and primary
input availabilities may be investigated in the model.
However, the linear programming formulation retains the assumptions of hori-
zontal supply functions (up to the point where capacity is reached) and vertical final
demand functions for each sector as well as fixed proportion production functions.
Hence, the demand for commodities and supply of factors are assumed to remain
constant no matter what happens to prices. In the linear programming framework it is
natural to interpret the shadow prices that result as a by-product of the solution as
equilibrium prices. However, these prices cannot be interpreted as market-clearing
prices of general equilibrium theory because endogenous prices and general equilib-
rium interaction to simulate competitive market behaviour cannot be achieved using
the linear programming specification. Thus, by using a linear programming formula-
tion, without representing a realistic price system in which endogenous price and
quantity variables are allowed to interact, the interplay of market forces cannot be
described properly. These are simplifying assumptions which severely restrict the
usefulness of the linear programming formulation of the input–output model.
In linear programming problems, the solution is guaranteed to occur at one (or
more) of the vertices, of the feasible set. This implies that the optimal solutions are
always to be found at one of the extreme points of the feasible set, and the solution
will constitute a basic feasible solution of the linear programming problem. Conse-
quently, all we need is a method of determining the set of all extreme points, from
R. Noren, Equilibrium Models in an Applied Framework,Lecture Notes in Economics and Mathematical Systems 667,
DOI 10.1007/978-3-642-34994-2_3, # Springer-Verlag Berlin Heidelberg 2013
21
which an optimum solution can be selected.1 However, this constitutes a significant
drawback of the applicability of the model because the linear programming specifi-
cation restricts the field of choice to the set of extreme points. Unlike the points of
tangency in differential calculus, the extreme points are insensitive to small
changes in the parameters of the model. That reduces the attractiveness of the
model for comparative static experiments. In order to include some elements of
flexibility within the system and make the linear programming model more realis-
tic, it is desirable to allow for the inclusion of several resource constraints and to
work on a highly disaggregate level. On the other hand, this will substantially
increase the amount of data required to implement the model. A technique which
removes any of the short-comings mentioned above will greatly improve the
applicability of the model.
For this purpose a straightforward extension of the linear programming model,
incorporating demand by sector and factor supply functions, will be developed. From
a complete set of demand and factor supply functions with only the demand and factor
prices as endogenous variables, it is then possible to compute the set of prices and
quantities that determines an economic equilibrium. The incorporation of demand and
factor supply functions provides a more realistic description of the aggregate market
conditions faced by individual decision makers. The Harrington (1973) formulation
of the Takayama and Judge (1964a, 1964b, and 1971) quadratic programmingmodels
of spatial price equilibrium operate in this way and will be followed to provide a
linear activity model for modelling economic equilibrium. This approach represents a
structure, where the technological data and estimates required to implement the
problem are to a great extent compatible with traditional linear programming models.
3.1 The Quadratic Programming Problem
In the quadratic programming formulation of the linear activity model both the
prices and quantities are determined endogenously within the model. In an
optimisation approach, the model is formulated in terms of the maximisation of
the sum of consumers’ and producers’ surplus.2 Based on empirically generated
demand and supply relations, this formulation of the objective function is used to
replace the utility and welfare functions of conventional economic theory.
Given downward sloping final demand and upward sloping factor supply curves,
relative price changes occur between sectors. Constraints on the model’s solution in
the form of fixed proportion production functions, current capacities and primary
resource availability are retained. Given this specification, the existence of a two
way feed-back in which quantity can influence price and price can influence
quantity for each sector, is developed.
1 The simplex method of linear programming represents such a method.2 See Noren (1987). The numerical tables are also presented in Noren (1991).
22 3 The Planner and the Market: The Takayama Judge Activity Model
The feasible set for quadratic programming problems is completely similar to the
feasible set for linear programming problems. On the other hand, the optimum value
of the objective function might occur anywhere in the feasible set. An optimum
solution may be on the boundary on the constraint region, but not necessarily at a
vertex or an extreme point, as we would expect in linear programming. Hence, the
quadratic programming model must permit consideration of non-basic solutions.3
Consequently, the field of choice extends over the entire feasible set and not merely
the set of its extreme points. In contrast to the linear programming model, we do not
have to work with a highly disaggregated model to increase the number of the
extreme points, and hence, extend the field of choice in the economic model. In the
quadratic programming formulation of the linear activity model, a framework has
been developed, that firstly, attempts to capture the role of prices and the workings of
a competitive market system, and secondly, the solution is not necessarily an extreme
point. The latter property implies that the solution is not so insensitive to small
changes in the parameters of the model. In fact, two of the major shortcomings of the
linear programming model have been overcome.
The theoretical basis of the model that will be presented in this chapter was
outlined in 1952 when Samuelson pointed out that an objective function whose
maximisation guarantees fulfilment of the conditions of a competitive market exists.
Samuelson defined this function as the “net social payoff” to avoid any association
with conventional economic concepts. Samuelson was the first to mention the possi-
bility of maximising the sum of consumers’ and producers’ surpluses to compute a
competitive equilibrium through an optimising model by showing how the problem of
partial equilibrium within spatially separated markets, as formulated by Enke (1951),
could be solved through mathematical programming. In the 1964 papers, Takayama
and Judge using linear price dependent demand and supply functions to define an
empirically oriented “quasi-welfare function”, and hence, extended the Samuelson
formulation so that the spatial structure of prices, production, allocation and con-
sumption for all commodities could be determined endogenously within the model
with quadratic programming. This work was followed by articles by Plessner and
Heady (1965), Yaron et al. (1965), and Plessner (1967), which contributed to the
formulation of the quadratic programmingmodel. In the development of the quadratic
input–output model, Plessner’s (1965) formulation of the Walras-Cassel model as
a quadratic programming problem has been of particular methodological interest.
Harrington (1973) followed the contribution of Plessner by showing how an
input–output model can be solved as a quadratic programming model, hence the
quadratic input–output model. The resulting quadratic input–output model is a theore-
tical improvement over the Leontief input–output model by the direct inclusion of
the pricing mechanism endogenously in the model. Thus, the methodological contri-
bution is the incorporation of the pricing mechanism in the programming model.
3 The main disadvantage of most quadratic programming algorithms is the large number of
calculations required for convergence to a solution. This implies that the quadratic programming
formulation is considerably more difficult to solve numerically than the linear programming
model.
3.1 The Quadratic Programming Problem 23
The model is a linearised version of the Walras-Cassel general equilibrium model
(linearised factor supply and commodity demand functions) which utilises the basic
Leontief input–output structure as a production relationship. Given the linearised
factor supply and commodity demand functions, both the prices and quantities are
determined endogenously. In technical terms, the shadowprices are incorporated in the
objective function. The solution of the quadratic programming problem can be
characterised as a simulation of market behaviour under the assumption of
competition.
The quadratic programming model presented in this chapter is applied for the
evaluation of the pattern of domestic production and trade of the Swedish economy.
The evaluation of the pattern of comparative advantages of the Swedish economy is
carried out as an analysis of the choice between import and domestic production in
a temporary equilibrium framework with exogenously given world market prices,
exports and domestic production capacities.
3.2 Specification of the Model
In developing the model, Hotelling’s (1932) total benefit function, based on empir-
ically generated demand and supply relations, is used to replace the utility and
welfare functions of conventional economic theory. We assume aWalrasian system
of private expenditures and factor supply functions, where the demand and supply
quantities are given as linear functions of the commodity price pj and factor price wh
respectively. Given this specification, we treat the aggregate demand and factor
supply functions as if they could be generated by a single representative individual.
To incorporate price-dependent demand and supply functions and derive an
economic equilibrium, mathematical models can be formulated with an objective
of maximising the sum of consumers’ plus producers’ surplus. Consumers’ plus
producers’ surplus or net social benefit is measured as the area between the
compensated demand and factor supply curves (after adjustment to remove income
effects) to the left of their intersection. The most obvious reason for the use of this
objective function is that its behavioural implications are consistent with theoretical
economic behaviour of the participants by sector. An important, although obvious
point, is that sector commodity supply curves and factor demand curves are not
required as they are already accounted for in the system by the fixed factor
proportion production functions calculated from the input–output table.
When this objective function is maximised, subject to the fixed proportion pro-
duction functions, a perfectly competitive equilibrium solution results.4 Constraints
reflecting the production capacities of the production sectors may alter the result, but
in a manner which continues to maximise producers’ and consumers’ surplus. Thus,
4 Takayama and Judge (1964a) present an existence proof based specifically on a mathematical
programming model of a space-less economy. This proof establishes the existence of a perfectly
competitive equilibrium in a mathematical programming framework of the general equilibrium of
an economy.
24 3 The Planner and the Market: The Takayama Judge Activity Model
the market is viewed as a mechanism for maximising the sum of producers’ and
consumers’ surplus. In technical terms, the shadow prices are incorporated in the
objective function. Hence, the solution of the quadratic programming problem can be
characterised as a simulation of market behaviour under the assumption of competi-
tion. Within the competitive framework, it is assumed that each domestic production
sector and the individual groups of consumers are composed of many competitive
micro units, none of which can individually influence quantity or commodity price.5
The concept of consumer’s surplus is defined as the difference between the
maximum amount the consumer would be willing to pay for the commodity and
what he actually does pay for it.6 In equilibrium, the consumption of the i:thconsumer is at the level at which the willingness to pay for the last consumed
unit is equal to its price.
The factor supply curve is upward sloping and measures the marginal cost of the
factor specific to the sector. Diagrammatically, the producer’s surplus is measured
as the area below the price and above the factor supply curve.7 This area has to be
identified with what Marshall (1925) called quasi-rent. Marshallian quasi-rent is
defined as the excess of the price over the marginal cost of the factor (labour) which
accrues to the producer or the factor owner as a profit in the short-run. Within the
short period, during which capital retains its sector specific form and the other
factor is fixed in price, the area above the supply curve as a measure of quasi-rent is
clearly relevant. Quasi-rents generally arise either because it takes time for new
firms to enter or because certain factor prices may be fixed over the short-run.
Generally, the term producer’s surplus is somewhat misleading, because it does not
identify which particular factor, and hence, factor owner to whom the rents are to be
imputed.8 Anyhow, economic rent can be defined to provide a measure of the
welfare change arising from a movement of factor prices, commodity prices
being constant; in exactly the same way that consumer’s surplus provides a measure
of the welfare change arising from a movement in commodity prices, factor prices
being constant.
In order to manage this problem computationally, we assume that linear
functions are acceptable approximations for the private consumption and factor
5 In this context the artificial nature of the objective function must be emphasised. As Samuelson
(1952) noted “This magnitude (the objective function) is artificial in the sense that no competitor in
the market will be aware of or concerned with it. It is artificial in the sense, that after an invisible
hand has led us to its maximisation, we need not necessarily attach any social welfare significance
to the result” (p. 288).6More rigorously, the difference between the money value of the total utility of the consumer’s
purchase and the money he actually pays for it.7 Strictly speaking, the producer’s surplus is the difference between total revenue from his sales,
minus the area under his marginal cost curve.8 Under perfect competition, the producers’ surplus is captured by the factor owner (owners of
specific capital equipment) in form of rent. In this model all the rents must be paid to the
households. Thus, it is possible to have a producers’ surplus and yet zero profit in competitive
equilibrium.
3.2 Specification of the Model 25
supply functions. This specification results in a quadratic net-benefit or, in the
terminology of Takayama and Judge, quasi-welfare function, and market equilib-
rium may therefore be computed by the techniques of quadratic programming to
obtain the optimum prices and quantities.
The final demand and factor supply functions are specified by the Cassel-Wald
(1951) specification, i.e. demand and factor supply functions are functions of
demand respective factor supply prices alone. As demonstrated by Harrington
(1973) the demand and factor supply functions specify, together with the
specifications of the industry supply system, a consistent system without loss of
generality of the Dorfman et al. (1958) specification of the Walras-Cassel model of
a perfectly competitive economy.
To understand the nature of the programming formulation,9 let the consumption
(private consumption) of the final commodity xj be a linear function of price such that:
xj ¼ γj � pj Σi; νij (3.1)
where we assume γj > 0 and νij > 0 for all j > 0. xj is the quantity of demand of the
desired commodity j, pj is the price of the sector’s product, γj is the intercept term,
the νij represents the slope coefficient. Note that the demand function is independent
of the sector activity, i.e. the income variable is dropped from the demand func-
tion.10 Alter-natively, the inverse of the demand-quantity function11 above is the
demand-price function:
pj ¼ αj � Σi; ωij xij (3.2)
Where we, as for Eq. 3.1, assume αj > 0 and ωij > 0 for all j > 0. αj is the
intercept term, ωji represents the slope coefficient and xij the i:th consumer’s
demand of the desired commodity. The matrix of slope coefficients is assumed to
be symmetric and positive definite for all j. The demand functions are continuous,
differentiable and monotonically decreasing functions of the consumed quantity xj,i.e. ∂(Dj((xj))/∂xj < 0 for all j > 0. The adjustment of prices according to the
9A general survey of techniques for formulation and solving multimarket general equilibrium
models in the mathematical programming framework have been spelled out in detail by Takayama
and Judge (1971).10 This formulation does not incorporate the income generated by the sector as a simultaneous
shifter of the model’s commodity demand function. If the sector under consideration is small
relative to the entire economy, this should not be a serious problem. However, if a major sector or
set of sectors is of interest the income generated within that sector (or sectors) may have a major
impact on aggregated consumer demand.11 In making the model operational, inverted demand and supply functions are applied. The
inversion simplifies the mathematical exposition of the model and the interpretation of the
solutions rather than the direct demand and supply functions. Dorfman, Samuelson and Solow
claim that this inversion is not admissible (Dorfman et al. 1958, p. 352). However, their argument
does not apply to the linearised Walras-Cassel model.
26 3 The Planner and the Market: The Takayama Judge Activity Model
market means that the pj’s may be regarded as functions of the xj’s, in spite of
individual consumers considering the pj’s fixed.The area under these demand curves and above the price represent consumers
surplus for each desired commodity. Integrating the set of the demand curves to
determine the area under the curves, a market-oriented net benefit function, denoted
by W, for the economy (comprising all desired commodities) may be specified as a
strictly concave quadratic function:
Wðx�Þ �ðx�0
Σj
αj �Xi
ωijxij
!dxj (3.3)
Where x* is a vector. Given the specification above, ωij � ωj. Hence:
Σi; ωjixij ¼ ωjΣi; xij ¼ ωjxj (3.4)
This results in:
Wðx�Þ �ðx�0
Σjαj � ωjxj� �
dxj (3.5)
Dropping the superscript, we obtain:
WðxÞ � Σj; αjxj�1=2Σj; ωjx2j (3.6)
More compactly, the function (3.6) may be written as:
WðxÞ � α0x�1=2x0Ωx (3.7)
where the matrix of slope coefficients is a diagonal, with zeros as off-diagonal
elements.
Similarly, we assume that the supply of factor quantities rih (primary commodities)
depends on the market prices of its productive services. Hence, let the inverse factor
supply function of commodity h (rih the supplied quantity of the primary commodity
h owned by the i:th consumer) be given by:
wh ¼ βh þ Σi; ηihrih (3.8)
Where we usually assume βh > 0 and ηih > 0 for all h > 0. wh is the price of the
primary commodity h. rh is the supplied amount of the primary commodity h. βh isthe intercept term and ηih represents the slope coefficient. The matrix of slope
coefficients is assumed to be symmetric and positive definite for all h. The supplyfunctions are continuous, differentiable and monotonically increasing functions of
the supplied quantity rh, that is ∂(Sh((rh))/∂rh > 0 for all h > 0.
3.2 Specification of the Model 27
The area under the factor supply curves (comprising all factor supply curves) is
total cost and may mathematically be written as:
Wðr�Þ �ðr�0
Σh
βh þXi
ηihrih
!drh (3.9)
According to the specifications above, we have here a model which will simul-
taneously determine the market demand price on final commodities (consumed
quantities of xj) together with the input market equilibrium prices on its primary
commodities (factor supplies of rh).The sum of producers’ and consumers’ surplus is then found by computing the
difference between the area under the final demand curves and the area under the
factor supply curves.
Wðx; rÞ �ðx�0
Σjðαj � ωjxjÞdxj �
ðr�0
Σhðβh þ ηhrhÞdrh (3.10)
Thus, total net benefit (comprising all desired commodities and all factor supply
curves) for the stipulated economy is the line integral of individual demand and factor
supply relations of which consumer’s and producer’s surplus is a part. The model can
actually be looked on as combining Koopmans (1957) linear production model with
Walras’s conception of the market, in a quadratic programming formulation.
The matrix of substitution terms in the demand and factor supply functions must
be symmetric. These conditions are the so called integrability conditions. They playan important role in the formulation of the model. The integration process is known
to be feasible when certain symmetry conditions are satisfied by the functions being
integrated, provided that these functions are sufficiently smooth. Hence, the sym-
metry conditions are often simply called the integrability conditions. Given the
symmetry conditions, a utility and cost function exists from which a consistent
demand respective supply function can be derived.12
If the substitution termmatrices do not conform to the assumption of symmetry the
integrability conditions are not satisfied, then we are unable to construct the net
benefit function given above. From an application standpoint, this presents
difficulties. However, the implications of this requirement vary depending upon
whether we are concerned with supply or demand. The classical assumptions of the
theory of production yield the symmetry conditions of the supply functions (Zusman
1969). Takayama and Judge (1971) have pointed out that if the integrability
conditions do not hold, then the system is still solvable and interpretable in terms of
net social monetary gain which is defined as total social revenue minus total social
12 For details, see Varian (1984), pp. 135–139.
28 3 The Planner and the Market: The Takayama Judge Activity Model
production cost. Only the connection to utility maximisation and cost minimisation is
lost by violation of the integrability conditions, not the solvability of the system.13
The symmetric condition is a necessary and sufficient condition for what is known
as path-independence. This implies that the cross-price effects (compensated) are
equal over all commodity pairs. In the present context, this means simply that in
whatever way the order of price changes is calculated the adopted measure of
consumer’s and producer’s surplus for the combination of these price changes is
uniquely determined. The symmetry of the substitution termmatrices (Slutsky terms)
is exactly the condition under which the integral W(x,r) is solely dependent on the
terminal price vectors, and thus, regardless of the order in which the price changes are
taken, i.e. independent of the path. However, given a demand function including the
income variable, the path-independence condition requires that the income
elasticity’s are identical across all commodities of interest. Given the property that
the weighted sum of the income elasticity’s, where the weights are the shares of
income spent on each commodity, sums to one, all income elasticity’s are equal, and
thus, equal to one.14 Unitary income elasticity’s are the demand functions derived
from homothetic indifference maps. This implies that all Engel curves are straight
lines through the origin, i.e. at all income levels, a constant proportion of total
expenditures is allocated to each commodity.
3.2.1 The Introduction of Foreign Trade
Most commodities can be supplied not only by domestic production, but also by
importation. A standard approach is to specify imports as an alternative source of
supply of commodities classified by the input–output sectors (Technically as an
alternative column in the input–output table). A different approach is to specify
imports as a primary input that is not produced in the economy (Technically as a
row in the input–output table).
In the first approach, imports are specified as competitive, here denoted Mj,commodities which can be produced within the country but which are, as an
alternative to domestic production, also imported. The imported commodity is
here viewed as a perfect substitute for the domestically produced commodity.
Consequently, those imported commodities which the agents are free to select for
domestic production are classified as competitive imports. In this context, any
particular commodity classified as competitive imports is assumed to be tradable
in the international market, and has identical characteristics, whether it is produced
at home or abroad. Formally, competitive imports are treated as if they were
13 Takayama and Judge (1971), pp. 121–126 and pp. 233–257.14 The path-independence condition is also fully satisfied if the income elasticity’s of demand of all
commodities are zero (McCarl and Spreen 1980). In this model the income variable is dropped
from the demand function. Thus, the path-independence condition is satisfied.
3.2 Specification of the Model 29
delivered to the corresponding domestic industries and then distributed by these
industries together with the domestically produced amounts. Thus, the inputs aijZjstate the sums of produced and imported amounts, and not merely the produced
amounts.15
In the second approach imports are specified as non-competitive, here
denoted mqjZj, and instead of perfect substitutes for domestic production, imports
are treated as a complementary input, completely different from domestically pro-
duced commodities. This type of imports consists of commodities which cannot be
produced within the country. Non-competitive imports including predominantly
those commodities which are technically infeasible, and commodities whose produc-
tion is economically unviable because of the present market situation compared with
their minimum scale of production. In our notation, mij denotes the input coefficient
of non-competitive imports and Zj the extent of which the process j is utilised.When a commodity is imported there is an outlay of foreign currency per unit of
imported amountMj respectivemijZj. If PW denotes the world market price in foreign
currency, �PWjMj and �PWjmijZj ex-press the outlay of foreign currency. On the
other hand, when a commodity is exported, denoted Ej, there is a receipt, expressedby PWjEj, of foreign currency earned per unit of exported amount Ej. Consequently,foreign currency is here an intermediate commodity, where the import process
requires foreign currency as input, and foreign currency is the output of the export
process. Thus, in this context there are also given resources, but of foreign currency
only. These resources are made up of net export earnings plus net foreign capital
inflow, denoted F. In this model the amount of net foreign capital inflow is assumed
exogenous. Given the exchange rate, denoted ER, it follows that foreign trade can bedescribed as to be carried out by means of processes with fixed relations. Compatible
with the assumption made for domestic production, it will be assumed that an import
process involves importation of one single commodity. This assumption re-places, as
for domestic production, an optimisation requirement.16 Consequently, we also
assume that an export process leads to the export of one commodity only.
The effects of transportation costs and tariffs are taken into consideration by
including transport costs and tariffs into import prices (tariff augmented world
market prices). Hence, the currency spent on importing a unit of a commodity is
generally somewhat larger than the amount earned by exporting it.17 If it were
smaller, this would mean that the price in the exporting country would exceed the
price in the importing country, which is not compatible with interregional general
equilibrium. In this model world market prices of traded commodities are assumed
to be given. The assumption of given world market prices (the small country
15 The exposition in this section is based on and similar to that of Werin (1965).16 Optimisation implies that the import process, given the smallest currency outlay, as well as the
production process, given the best technique available, is chosen.17 Statistically, imports are calculated in c.i.f. prices and exports in f.o.b. prices. Given this
specification, the currency outlay for imports will not be proportional to the existing world market
prices. This implies that the foreign exchange constraint will not correctly reflect the conditions
prevailing on the world market.
30 3 The Planner and the Market: The Takayama Judge Activity Model
assumption) implies that the country is confronted with infinitely elastic demand for
its exports and supply of its imports, so what the level as well as the pattern of
imports and exports may be endogenously determined only subject to the foreign
exchange restriction.
Considering the assumptions made, the production system is re-presented by an
input–output model extended to include foreign trade as an alternative to domestic
production. Each commodity can now in principle be supplied by two different
activities. One of them is the production activity, the other the import activity,
which is the result of the outlay of foreign currency. This means substitution
possibilities between inputs for the supply of various commodities. A linear activity
model which takes foreign trade into account is, in certain respects, quite similar to
a neoclassical model.18
The foreign exchange constraint (Eq. 3.11) restricts the amount of foreign
currency that can be spent on imports. The supply of foreign currency is generated
through exports and net capital inflows. PWj denote the world market price of each
commodity classified by the input–output sectors. In this model, imports will betreated both as an alternative (and identical) source of supply of commodities
classified by the input–output sectors and as another input (composite) that is not
produced in the economy, analogous to capital and labour. Technically, competitive
imports are placed outside the inter-industry part of the input–output table, specified
by sector of origin, and non-competitive imports are kept within the inter-industry
part of the input–output table, specified by sector of destination.
Σj; Σi; PWjmijZj þ Σj; PWjMj � Σj; PWjEj þ F (3.11)
3.3 The Programming Formulation
Given the net benefit function, and the constraint set as specified above the problem
takes the following form, i.e. maximise:
Wðx; rÞ � Σj; αjxj � 1=2Σj; ωjx2j � Σh; βhrh � 1=2 Σh; ηhrh
2 (3.12)
Subject to
Zj þ Σj; mijZj þMj � Ej � Σj; aijZj � ΣiDij (3.13)
Σj; bhj Zj � Σi; rih (3.14)
18However, if the model does not include any further restrictions on exports and imports, the
assumption of constant returns of scale in production together with endogenous choice in trade
may lead to an unrealistic specialisation in either trade or domestic production.
3.3 The Programming Formulation 31
cij Zj � Kij (3.15)
Σj; Σi; PWjmijZj þ Σj; PWjMj � Σj; PWjEj þ F (3.16)
Zj � 0; Mj � 0; Ej � 0; Dij � 0; rih � 0; Kij � 0
Making use of the Kuhn-Tucker conditions, the necessary conditions which must
hold for the optimum xoij; roih; Zoj ; Mo
j ; poj ; woh; voij; ERo to be a non-negative saddle
point of the Lagrangean, are:
@Lo
@xij¼ αj � ωjx
oij � poj � 0
�00 � < 0 ) xoij ¼ 0 ð3:17Þ
@Lo
@rih¼ �βh � ηhr
oh þ wo
h � 0
�00 �< 0 ) roih ¼ 0 ð3:18Þ
The constraints of the domestic activities will be the same as in the linear
version. See the discussion in Chap. 2, Sect. 2.5. However, the inclusion of foreign
trade implies two other constraints in the quadratic model. The new constraints are
discussed below as constraint (3.23) and (3.24).
For a given vector of pre-equilibrium prices pj and wh, these prices are revised
until the shadow prices poj and woh associated with the commodity balance Eqs. 3.13
and 3.14. If so, the solution is an equilibrium solution. Thus, the dual variables from
Eqs. 3.13 and 3.14 equals the maximum price the consumers are willing to pay for
the consumption of the commodities available to them, and the minimum price at
which they are willing to supply labour service from their initial endowment of
leisure. If not, the demand and supply prices are revised and start a new function
evaluation. In this way shadow prices have a feedback effect on the demand and
supply prices specified in the objective function. As stipulated above, this is what
leads to the similarity between the market mechanism and the optimisation formu-
lation of the model. A planning authority can use the shadow prices generated by
the plan to decentralise decisions because they are signals of relative scarcity of the
constraint to which they are attached. However, when imposing a number of
additional ad-hoc constraints to make the solution more realistic, the constraints
result in distortions in the shadow price system. If such constraints can be justified
as additional system constraints that define a reasonable notion of economic
equilibrium, there is no theoretical problem to interpret the solution as reflecting
the operation of a market system (Taylor 1975).
Starting with the shadow demand price, denotedpoj ,when the consumption of the j:
th commodity is positive, must exactly be equal to the demand price pj, the maximum
price the consumers are willing to pay for the consumption of the quantity of the
32 3 The Planner and the Market: The Takayama Judge Activity Model
commodity xj, which in turn are generated by the optimum demand quantity xoj .
However, if xoj ¼ 0, the shadow demand price is greater than or equal to the demand
price. Thus:
if xoj > 0; then αj � ωjxoj ¼ poj ð� 0Þ; (3.19)
if xoj ¼ 0; then αj � ωjxoj � poj ð� 0Þ; (3.20)
for all j.The factor supply equilibrium stipulates, that when the optimum supply quantity
of the h primary commodity is positive, the shadow supply pricewoh must exactly be
equal to the supply price (factor cost) wh, the minimum price at which the resource
owners (consumers) are willing to supply rh, where roh are generated by the optimal
supply quantities roh. However, if roh ¼ 0, the shadow supply price is less or equal to
the supply price. Thus:
if roh > 0; then βh þ ηhroh ¼ wo
hð� 0Þ; (3.21)
if roh ¼ 0; then βh þ ηhroh � wo
hð� 0Þ; (3.22)
for all h.The individual country becomes a price taker in the small open economy model,
because the world market prices of traded commodities are assumed to be determined
in the international market. The domestic economy will at the optimum adjust to the
relative world market price ratio. In a free trade economy,19 the direction of trade will
be determined by the requirement of equality between the domestic and the world
market price ratio. It is the difference between these ratios that leads to trade. Thus,
efficiency requires equality among world market prices, domestic prices, and produc-
tion costs. Since the world market prices are assumed to be given, these prices
determine the domestic shadow prices of tradables.
@Lo
@Mj¼ poj � ERoPWj � 0
�00 �< 0 ) Moj ¼ 0 ð3:23Þ
Next condition (3.23), relates to the alternative way of supplying a commodity,
namely by importation. Condition (3.23) state, that when the optimum import
19 Using the small-country assumption and also assuming that domestically produced and imported
commodities are perfect substitutes this specification leads to extreme specialisation in either trade or
domestic production whenever there are no established domestic capacity constraints. The sector-
specific capacity constraints in this model are used to limit this problem. This implies that the
domestic shadow price system is no longer a simple reflection of world market prices.
3.3 The Programming Formulation 33
activity Moj is positive, the shadow price poj of the imported commodity must be
exactly equal to the value (cost) of the outlay of foreign currency. If the shadow
price poj is lower than the imputed cost of importing the commodity no importation
of the commodity will take place. Production will expand until domestic production
costs rise to the world market price level, converted into a domestic price by the
shadow exchange rate ERo. Consequently, as long as domestic production costs are
lower than established world market prices, it will be profitable to expand domestic
production for exports. On the other hand, if the domestic price is greater than the
world market price, the commodity will not be produced. If the country can always
import at a cost of poj it is never optimal to produce at a marginal domestic cost
higher than poj . This leads to excess domestic capacity which is reflected by a
shadow price of zero for installed capacity. Since, our model only contains
tradables; the shadow exchange rate is simply defined as a conversion factor from
foreign exchange units to domestic commodity units, and has no significance in
terms of relative domestic prices.20
Finally, condition (3.24) below state, that if the optimum price of foreign exchange
is positive, the foreign exchange equilibrium requirement for the economy is exactly
met. Note, that for any positive activity the shadow exchange rate ERo can never be
zero because it is always possible to use foreign exchange to purchase commodities
from abroad.21 If the shadow price of foreign currency is zero at the optimum no
activity (production and importation) take place in the domestic economy. Given this
specification, there is the assumption of a flexible exchange rate system, in which
exchange rate adjusts continuously so as to maintain the foreign exchange constraint
in equilibrium.22 However, specifying tariffs on currency outlay for imports implies
that the domestic shadow prices would reflect the existing tariff structure, and the
tariff-ridden domestic market prices will not be proportional to the existing world
market prices. Hence, the foreign exchange constraint will not correctly reflect the
conditions prevailing on the world market.
@Lo
@ER¼ Σj;PWjE
oj þ F� Σi; Σj;PWjmijZ
oj � Σj;PWjM
oj � 0
�00 �> 0 ) ERo ¼ 0 ð3:24Þ
In the closed economy the basic technological and demand variables determine the
domestic shadow price system.23 However, the situation is quite different in a free
20With non-tradables, the shadow price of foreign exchange will reflect the relative scarcity of
tradables with respect to non-tradables.21 For a discussion of this mechanism, see Dervis et al. (1982), pp. 75–77.22 Assuming given world market prices, an increase in domestic prices implies a depreciation of
home currency. Conversely, a decrease in domestic prices implies an appreciation of home
currency. See further, Sodersten (1980), pp. 315–328.23 The discussion that follows is based on Dervis et al. (1982).
34 3 The Planner and the Market: The Takayama Judge Activity Model
trade economy where the domestic market is small in relation to the world market.
Given the assumption of perfect substitutability between imported and domestically
produced commodities, the small-country assumption implies that the individual
country becomes a price taker facing exogenous world market prices. The theory of
international trade suggests that, as far as some commodities are actually imported or
exported, the domestic shadow prices among them tend to converge to their relative
world market prices.24 Consequently, world market prices determine the domestic
shadow prices of tradables, and a given commodity has (at equilibrium) the same
price whether it is imported or produced domestically. Hence, whereas supply and
demand determine domestic shadow prices in a closed economy, they will adjust to
world market prices in the small open economy.
3.4 A Temporary Equilibrium Specification
The static model as presented above has no formal link between capital formation
and production capacity. Capital commodities are assumed exogenous without any
correspondence to the effect that is created by the supply of investment from sectors
producing capital commodities (investment in final demand). However, a tempo-
rary equilibrium specification endogenises investment and considerably extends the
requirement of consistency in the model. The period output of the capital stock
requirement is inserted as a predetermined variable for the next period optimiza-
tion.25 Once capital stock requirement by sector of destination is established, its
sectoral allocation into a demand for investment commodities by sector of origin
must be specified.
Operationally, the solution for each period is used to create the next period’s
model parameters. Thus, the model is of the temporary equilibrium type. It will
solve the market for equilibrium prices and quantities for one period and then add
the solution obtained to the predetermined variables that are needed to obtain the
market equilibrium solution for the next period. The model does not take into
account future markets despite the fact it explicitly considers time. There is no
inter-temporal optimization26 and the agents have no expectations about future
prices. This concept of equilibrium as static and temporary implies that we are more
interested in the outcomes of the adjustment that yields a new temporary static
equilibrium position than in the dynamics of the adjustment process itself.27
24 Differences may exist due to transportation costs and tariff rates.25 Given the specification of the model, also private consumption is inserted as a pre-determined
variable for the next period optimization.26 In intertemporal models, agents have rational expectations and future markets are considered
when optimizing. Endogenous variables follow an optimal path over time and there are no
incentives to deviate from this path at any point of time.27 Hence, we can overlook the issue of adjustment.
3.4 A Temporary Equilibrium Specification 35
Investment is made up of two parts, replacement investment and net investment.
Replacement investment is that portion of the total which exactly maintains the
capital stocks while net investment is that portion which depends on the level of
demand. In this specification, only net (private) investment in buildings and
machinery is considered. Logically, we disregard depreciation. Another component
of capital formation is inventories. However, the model treats inventories as an
exogenously given component of final demand, and thus, does not incorporate
inventories in the investment concept.
The change in capital stock is by definition the amount of investment. As long as
domestic demand is unchanged, the capital stock is adequate and no investment is
needed. Increases in domestic demand, however, call for additional capital and net
investment is positive. Formally, we assume investment (given the assumption of
full capacity) to be linearly dependent on the current period’s request for newcapacity. This implies that investment adjusts immediately to changes in capacity
requirement within a single period.28 Nevertheless, there is certainly reason to
suspect that in the real world firms do not respond immediately. Hence, it is
assumed that each period is long enough for relative prices to adjust to clear
markets. In quantitative terms, the request for capital commodities by sector of
destination ΔKj is translated into a demand for investment commodities by sector of
origin Ik (producing sectors of capital commodities). Thus we have
IiðtÞ ¼ ΣjτkjΔKjðtÞ (3.25)
Where τkj denotes the matrix of sectoral investment allocation shares, i.e. the
proportion of capital stock in sector j originating in sector k. Note that Στij ¼ 1 for
all j (summation is taken over i). The matrix of sectoral investment shares is
compiled by the Ministry of Finance for the 1984 Medium Term Survey Model
of the Swedish economy.29
It is important to note that the model, in this version, only considers positive net
investments. In other words, given a decrease in the capital stock requirement by
sectors of destination (ΔK < 0) the net investments by sectors of origin are zero.
For this alternative, only sectoral capital stocks are adjusted (scrapping) for the next
period optimization. Moreover, fixed coefficients are used to allocate investment
among sectors. Thus, profitability across sectors is assumed fixed over time. This
implies that we have no allocation process explicitly modelled, in which investment
gradually adjust to equalize profitability across sectors. Hence, the workings of
financial markets in the investment allocation process are ignored. Technically, the
capital stock in each sector is a well defined aggregate of various commodities with
28 This is the famous accelerator principle. In its simplest form, the accelerator rest upon the
assumption that the firm or industry at each level of distribution seeks to maintain its optimal
capital stock at some constant ratio to sales.29 SOU 1984:7, LU 84 (The 1984 Medium Term Survey of the Swedish Economy), Appendix 17,
Table 2:18. Only 9 sectors produce investment commodities for domestic capacity expansion.
36 3 The Planner and the Market: The Takayama Judge Activity Model
a fixed compositional structure (by sector of origin). Finally, there are assumed to
be fixed incremental capital-output ratio by sectors.30
3.5 Empirical Findings: Applications
As stipulated above, the model works stepwise from period to period, and solves the
market for prices and quantities. The solution for each period (four periods in total)
is used to create the next period’s model parameters. Hence, a sequence of
equilibria can be achieved. The period output of capital stock requirement, invest-
ment demand and private consumption are inserted as predetermined variables for
the next period optimization.31
The point of departure for the experiments below (here named applications)
is the version of the model which describes the techno-logical conditions,
labour costs, capacities and estimated demand relations representing the Swedish
economic situation in the year 1980 (benchmark equilibrium data set).32 This year
is selected since it con-forms with data availability, and capacity utilization during
the whole of 1980 on the average can be characterized as normal full capacity.
Thus, the 1980 data provide a comparative benchmark for the experiments in this
chapter. In all solutions, the same maximand is used, i.e. maximize the consumers´
surplus (Eq. 3.3), subject to the constraints (Eqs. 3.13, 3.14, 3.15, and 3.16). Given
the assumptions above, a foreign payments imbalance cannot arise. Moreover, we
assume that the labour constraint (Eq. 3.14) is binding, i.e. labour resources are used
to the maximum availability. In all solutions the total supply of labour resources is
given exogenously and assumed perfectly mobile and free to flow among all sectors
of the economy. Hence, labour moves across sectors until the value of its marginal
product is the same everywhere. This assumption, the value of that marginal
product of labour are equalized in all uses in equilibrium, permits labour payments
data by industry to be used as observations on physical quantities of labour in the
determination of parameters for the model.
In general terms, adjustment to structural equilibrium is a process where profit-
ability in the different sectors will adjust to a “normal” level of profitability for the
economy as a whole. For sectors where profitability is high relative to this normal
level, the adjustment to equilibrium implies an increase in domestic production
relative to other sectors. On the other hand, a sector where profitability is low
relative to the normal level, an adjustment to equilibrium implies a decrease in
30 The temporary equilibrium approach used in this study does not imply that the underlying
economic system is viewed as discrete. Instead, the discrete moments are simply approximations
(artificial to some extent) of the essentially continuous system being modelled.31 Adjustment costs for the installation of capital are not considered.32 The model of the Swedish economy comprises 24 sectors. These are defined in the Appendix, in
accordance with both the Standard Swedish Classification of Economic Activity (SNI) and the
code for the ADP system for the Swedish National Accounts (SNR).
3.5 Empirical Findings: Applications 37
domestic production relative to other sectors. Thus, a development which implies
that a country adjusts to its comparative advantages33 is characterized as an
adjustment towards equalizing the relative profitability between sectors. The results
of this adjustment are reflected in the direction of domestic production.
In technical terms, the domestic shadow prices adjust to the exogenous world
market prices in this model. Thus, the concept of a normal level of profitability for
the different sectors is determined by the relative world market prices. If the
domestic shadow price is greater than the world market price, the domestic produc-
tion of the commodity relative to other sectors will fall. If it is not possible to reduce
domestic production costs to the level of world prices, the commodity will be
imported altogether. On the other hand, if the domestic shadow price is lower
than the world market price, domestic production relative to other sectors will
expand at the expense of imports until domestic costs rise to the level of world
market prices.34 If this equality is not satisfied in the case when the adjustment to
equilibrium implies a zero import level, it would be profitable to expand domestic
production for exports.
Generally, due to the assumed linearity of the underlying technology, the
solution in the model imposes that fewer commodities will be produced domesti-
cally, but in increased quantities in the least-cost sectors. On the other hand, the
specialisation will lead to an increasing amount of import in the high-cost sectors.
In all experiments, it is the difference between the world market prices (here
assumed to be given)35 and the pre-trade domestic commodity transformation
rates that leads the model to take part in trade.
To obtain a reasonable pattern of specialisation, exports are assumed exogenous.
As exogenous values of exports we have maintained the 1980 figures. By this
assumption extreme specialisation is prevented. Unfortunately, these constraints
reduce the experimental attractiveness of the model.
Given the model specification, the equilibrium data of the former period provide
a comparative benchmark for each experiment (four experiments in total).36 Appli-
cation 1 is considered as the first period. It is important to emphasize that the results
have been obtained under strong simplifying assumptions. The results of the
experiments are presented in the Appendix 2 (Tables 3.3, 3.4, 3.5, and 3.6).
As a starting point for the experiments we assume an increase in the sectorally
fixed capital stocks by 10 %. This implies that domestic resources may be shifted to
33Given two sectors 1 and 2, the economy has a comparative advantage in sector 2 if the pre-trade
ratio of sector 2 costs to sector 1 costs is lower than the world price ratio.34 Following Norman (1983) a domestic sector is competitive if (and only if) its marginal cost is
lower or equal to its foreign competitor, measured in the same currency. To be compatible with the
concept of comparative advantage, and hence meaningful, marginal cost is here defined as long run
marginal cost. This implies that the concept of marginal cost includes payment to factors that are
fixed in the short run, e.g. capital.35 The world market prices are specified as unity prices.36 The first experiment (application 1) provides the benchmark data for the second experiment
(application 2) and application 2 provides the benchmark data for the third experiment (application 3).
38 3 The Planner and the Market: The Takayama Judge Activity Model
the lowest-cost sectors (given the capacity restriction) and thus increase the effi-
ciency in resource allocation. Logically, the model chooses to import in some
sectors (Sector definitions in Appendix 2, Table 3.2) rather than utilize the existing
capital stock. As expected, we obtain an increase in engineering (15) and a total
contraction of the shipyards (16). Moreover, the result obtained shows a decline of
domestic production in the basic metal industries (14). As specified above,
the sectoral demand for capacity expansion, evaluated in the former period (appli-
cation 1), is translated into investment by producing sectors in the current period
(application 2). In this connection, the increase in some sectors of the index
representing sectoral demand prices should be noted. The demand prices
(Tables 3.3, 3.4, 3.5, and 3.6, column 13) of the private consumption variables
are expressed in terms of an initially established index, assigned as 1,000. The
explanation for this increase in demand prices is that investment required for
capacity expansion (given as input from application 1) have increased for most
sectors producing capital commodities. Consequently, in some sectors a decrease
(crowding out) of other demand components (here, private consumption only) is
necessary to make capacity expansion possible. At the beginning (application 1 and
application 2) the request for capacity expansion is considerable. However, a
continuing fall in mobility, due to the limited supply of labour resources (measured
in terms of wages), increasing capital stocks in the investment sectors, and the
linear specification of the model, will in the long run reduce the demand for
capacity. The diminishing welfare effect, due to reduced potential in resource
allocation, is the main factor behind this development. Thus, in the next two
experiments (application 3 and 4) it is quite obvious that the demand for net
investment by sectors will fall. These calculations are presented in Table 3.1.
Capacity expansion and the process of structural transformation is restricted to
the existing structure of production. The technological structure is kept the same.
Not unexpected, the results presented in Table 3.1 indicate that the resource
transformation process alone is not sufficient to sustain a high rate of growth in
industrial real capital formation. Successively increasing investments in new tech-
nology, introduction of new commodities, and in its extension, the formation of
new activities (operations), are strongly needed to maintain the capacity for indus-
trial renewal.
From an evolutionary theoretical point of view (Schumpeter is among the classics
in this field) the model, and theory,37 outlined here is in this respect inadequate to
capture the process of structural renewal, and hence, the specification of the
mechanisms that creates incentives for the entrepreneur to enforce new investments
to maintain the capacity for growth. In assessing these results it must be emphasized
that investment is restricted to capacity expansion, i.e. net investment. Moreover, all
investments are in established industries and hence, according to the specification of
the model, directed to the production of a given set of commodities. In the real world,
37 The perfect competition theory defines the equilibrium state and not the process of adjustment.
(Kirzner 1973, p. 130).
3.5 Empirical Findings: Applications 39
however, investments made to increase the total capacity as well as the replacement
and scrapping of old production units change the production characteristics.
Investments in new capacity embodying best-practice techniques will decrease the
sector’s input coefficient at full capacity. Thus, new capacity has in general
input–output proportions different from those of existing production units due to
changed relative prices and technical progress, which may be embodied or
disembodied (learning by doing). Furthermore, investments introduce input–output
combinations, and in the long run, production of commodities which cannot be found
within the initial production possibility set.
Returning to application 4, the equilibrium model does no longer choose to
establish agriculture and fishing (1) and the mining and quarrying industry (3) in the
Swedish economy. On the other hand, engineering (15), wood, pulp and paper industry
(8) and chemical industry (11) belongs to sectors38 highly exposed to foreign compe-
tition, where expansion of domestic production is requested. Besides manufacturing,
private services (23) indicate an increasing share of domestic production.
In all experiments labour is assumed to be an aggregation of different skill
categories. In other words, labour is specified as homogenous in the model. Hence,
we can not value labour services (labour productivity) by skill group. Nevertheless,
the chemical industry and engineering are particularly intensive in terms of
technicians and skilled labour. In this respect, it seems that the joint utilisation of
human and physical capital provides an important input in the Swedish industry.39
In a model that does not include any restrictions on trade, a commodity is either
imported or exported, but never both.40 The explanation of this is that the commodity
imported and the commodity exported is assumed identical in the model. The
Table 3.1 Net private
investment by producing
sectors million kr – 1,975
prices
Sector Application Request in
1 2 3 4 5
1 281 326 358 0 0
2 346 684 0 0 0
7 161 296 203 100 87
8 807 500 334 155 175
15 21,840 34,862 26,841 11,405 6,253
16 1,299 2,251 2,476 0 0
17 28 20 22 0 0
19 26,118 15,978 12,562 3,697 1,904
23 650 3,214 2,490 967 324
38 The engineering industry is usually analyzed in terms of five sub-branches, i.e. metal goods
industry, machine industry, electrical industry, transport equipment (excl. ship-yards), and mea-
suring and controlling equipment industry. The machine industry is the largest sub-branch
(measured in number of employees and value added respectively). The sub-branches for metal
goods, electrical equipment and transport equipment are all roughly of the same size.39 See also Flam (1981), pp. 97–101.40 It is important to note that the level of aggregation will affect the value of the measures of intra-
industry trade. The higher the level of aggregation, the greater will be the share of intra-industry
40 3 The Planner and the Market: The Takayama Judge Activity Model
tendency for specialisation would be even more explicit, if we were to leave sectoral
capital stocks as endogenous variables.41 Needless to say, extreme specialisation in
production and trade conflicts with empirical evidence, which on the contrary, shows
relatively little specialisation on the sectoral level. However, as pointed out byWerin
(1965), the observed combination of domestic production and trade may be in
complete accordance with the theoretical model. First, the country under study
consists of many regions, which implies that a commodity may be imported to one
region and exported from another, but never be both imported to and exported from
one single region. Second, the same argument is applicable to the fact that the model
is specified to cover a period of some length. Hence, a commodity may be both
produced and traded at different points of time during the period of specification.
Finally, the commodities of the model are aggregates of different commodity
categories. For each of these commodities the theoretical requirement may be
fulfilled.
3.6 Comparative Advantages?
Whereas the Swedish economy, as expounded by the equilibrium experiments
above, tend to illustrate a comparative advantage in industries with large
requirements of human capital42 several empirical studies examine the net trade
patterns and the specialisation of production of Sweden with the EU (in the
beginning EEC) and other OECD countries, indicate a weaker market position in
human capital intensive industries (Ems 1988). Moreover, the R&D intensity did
not seem to influence the international competitiveness of the Swedish industry at
all. The pattern of change in the competitiveness of the Swedish industry versus the
EEC in 1970–1984 (Lundberg 1988) seem to reveal a comparative advantage in
industries requiring large inputs of physical capital and domestic natural resources.
Human capital intensity does not seem to have influenced net export ratios during
the period.
The discussion above has already stressed that a model that does not include any
restrictions on trade, a commodity is either imported or exported but never both.
However, during the post-war period there has been a marked increase in interna-
tional specialisation within the differentiated product groups and a substantial
trade (Grubel and Lloyd 1975). Although the share of intra-trade is reduced by disaggregation,
substantial two-way trade remains (Blattner 1977) on the most detailed aggregation level.41 A common approach to avoid unrealistic specialisation in multi-country trade models is to use
the Armington (1969) formulation, which treats similar commodities produced in different
countries as different commodities (commodity differentiation by country of origin). Bergman
(1986) makes use of the Armington formulation and applies a numerical solution technique in
order to solve the model.42 Nearly all available evidence indicates that Sweden has a comparative advantage in human
capital intensive production. A survey of these studies is given in Flam (1981), pp. 97–101.
3.6 Comparative Advantages? 41
growth in the share of intra-industry trade, i.e., imports and exports in the same
statistical commodity group.43 Thus, the increase in trade and specialisation is
dominated by reallocation on resources within rather than between industries.
The increase in intra-industry trade between Sweden and the EEC has been
particularly strong. Theoretical elements explaining the determinants of intra-
industry trade are based on the roles of product differentiation and economies of
scale. One point of departure in seeking to explain the growth of intra-industry trade
(Petersson 1984) has been the Lancaster (1980) theory which places central impor-
tance on product differentiation and scale economies specific to the product
(production runs). The adoption of a global production strategy and specialisation
within a limited range of commodities and product variants enables a country’s
producers to achieve long production runs. Similar opportunities for the producers
of other countries gave rise to a flow of import and an improvement in consumers’
choice. Hence, the existence of product differentiation (which is especially found in
consumer products) implies monopolistic competition which, from the consumer’s
viewpoint may correspond to a demand for variety in commodities.
Economies of scale with product differentiation normally prevails where
corporations make horizontal investments, i.e. to produce abroad the same lines
of commodities as they produce in the home market. Swedish firms which have
manufacturing affiliates abroad (multinational corporations) account for some
50 % of manufacturing employment in Sweden and almost 60 % of Swedish
exports (Swedenborg 1988). Moreover, they are dominating in engineering and
are highly internationalized. In 1986 less than 25 % of their total sales were sold in
the home market. Of the 75 % sold in foreign markets over half was produced
abroad. Empirical observation (Erixon 1988) suggest that the reduced market
share for Swedish exports may to a great extent be explained by the tendency for
Swedish multinational corporations to supply through local production in the
largest markets rather than through exports from Sweden. Thus, the size of the
market affects not only the volume of sales in a country but also leads to a higher
propensity to supply the market through local production (Krugman 1980).
Comparing the discussion above with the pattern of changes that emerges from
the experiments with the equilibrium model is interesting. In the equilibrium
model the necessary reallocation of sectoral resources is reached solely by an
adjustment in the structure of inter-industry trade. However, within industries
where the equilibrium experiments call for a substantial growth in domestic
production the economic gains is mainly intra-industry in nature. These
gains are in the form of economies of scale utilized to a great extent by
foreign production, rather than arising from reallocation of resources according
to comparative advantages. Thus, we have to be careful in interpreting the
obtained results in a too mechanical fashion.
43 The expansion of intra-industry trade in Europe which was particularly marked in the 1960s
appears to have largely halted in recent years. A somewhat similar situation is apparent for the US
(Hine 1988).
42 3 The Planner and the Market: The Takayama Judge Activity Model
3.7 Concluding Remarks
To conclude this chapter, it seems reasonable to compare the mathematical pro-
gramming (linear and quadratic) models above with models developed within the
tradition of computable general equilibrium (CGE) modelling. In such a compari-
son the programming models seem to be based on overly restrictive assumptions.
For example, while most standard CGE-models incorporate technology
descriptions that allow for factor substitution, there are fixed coefficients in the
linear programming model. Generally, due to the assumed linearity of the underly-
ing technology, the solution in the model imposes that fewer commodities will be
produced domestically, but in increased quantities in the least-cost sectors. On the
other hand, the specialisation will lead to an increasing amount of import in the
high-cost sectors. To obtain a reasonable pattern of specialisation, exports must be
specified to vary within certain limits or be assumed exogenous. By this assumption
extreme specialisation is prevented, but it is still a serious deviation from reality,
especially when foreign trade is a large part.
Another serious restrictive assumption is the treatment of maximising behaviour
by agents in mathematical programming models. In this chapter as well as in the
previous the central planner is assumed to be the only maximising actor. Theoreti-
cally, that conflicts with the market equilibrium price system, where the demand
and supply decisions are made separately and independently by various economic
actors. While most CGE-models incorporate complete systems of final demand
functions, usually derived from explicit utility functions, the demand representation
in the mathematical programming models are based on linear demand functions
with no explicit relation to utility maximisation under a budget constraint. Hence,
no ad hoc assumptions in order to avoid unrealistic solutions will be needed.
Not unexpected, these constraints reduce the experimental attractiveness of the
programming models in our study of a market economy.
Appendix 1: The Reformulation of the Walras-Cassel Model
To provide the methodology for the reformulation of the Walras-Cassel general
equilibrium model as a quadratic programming problem, and hence, the basic
structure of the quadratic input–output model, Harrington (1973) linearises the
Walras-Cassel model and specifies the Walrasian factor supply and commodity
demand functions into inverse form.44 The inversion simplifies the mathematical
exposition of the model while retaining the generality of the Walrasian factor supply
and commodity demand functions. Dorfman, Samuelson and Solow (1958) claim that
44 The Walras-Cassel model is specified in Dorfman, R., Samuelson, P. A. and Solow, R. M.,
(1958), pp. 346–389. The Walrasian model of the market system was first sketched by the
nineteenth-century French economist Leon Walras (1874–7).
Appendix 1: The Reformulation of the Walras-Cassel Model 43
this inversion is not admissible because there is no mathematical reason for assuming
the existence of inverse demand or supply relationships in a model were prices
depend on quantities only.45 However, their argument, as demonstrated by
Harrington, is well-founded in the general case but does not apply to the linearised
Walras-Cassel model. The quadratic input–output model is a linearised version of the
Walras-Cassel general equilibrium model which utilizes the inter-relatedness of
production established in the input–output structure. In this context, it is shown by
Harrington that the conventional input–output model is a limiting case of the
linearised Walras-Cassel model. In the linear form of the Walras-Cassel model the
assumptions of homogeneity of degree zero of factor supply and commodity demand
functions can be relaxed because the homogeneity constraint is satisfied elsewhere in
the model formulation. Furthermore, the Cassel-Wald specification of commodity
demand quantities as a function of product prices alone, and factor supply quantities
as a function of factor prices alone (Wald 1951), specify a consistent linear system
without loss of generality of the Walras-Cassel model.
In order to understand the underlying structure of the model that constitutes the
framework of this study a mathematical exposition of Harrington’s (1973) contri-
bution is given in this section.46 Let A denote a matrix of fixed coefficient produc-
tion processes, homogenous of degree one, partioned into a primary factor
transformation m � n matrix, Ar, and an intermediate commodity transformation
n � n matrix Aq. Let G(w, p) denote a linear factor market supply function defined
over all factor prices w (m1 1) and commodity prices p (n � 1), and let F(w, p)denote a linear commodity market demand function defined over all factor prices
w and commodity prices p.47 Thus, the assumptions above linearise the Walras-
Cassel model. Note, that the factor supply and commodity demand functions are not
assumed to be homogenous of degree zero in w and p.48 Under the assumption of
linearity of the factor supply and commodity demand functions the G and Fmatrices (Gr (m � m), Gq (m � n), Fr (n � m), Fq (n � n)) may be partitioned as:
Grwþ Gqp ¼ r and Frwþ Fqp ¼ q (3.26)
where q specifies a vector of final demand quantities, and r a vector of factor supplyquantities. Transforming factors into commodities require the following condition
on primary factor transformations:
45 Dorfman et al. (1958), p. 352 (footnote).46 The exposition in this section is based on Harrington’s own presentation of the subject.47 The factor supply functions are specified in the factor markets, the commodity demand functions
are specified in the commodity markets, and the transformation matrices are specified in the
production sectors.48 It is impossible to meet both the specification of linearity and homogeneity of degree zero in the
same function. Since F and G are matrices of constants they are by definition homogeneous of
degree one.
44 3 The Planner and the Market: The Takayama Judge Activity Model
Arz ¼ r (3.27)
Intermediate commodity transformations require:
½I� Aq�z ¼ q (3.28)
where z represents a vector of gross output per sector. [I � Aq] referred to as the
Leontief matrix, is based on the conditions of conventional input–output analysis,
hence, its inverse exists. Consequently:
½I� Aq��1q ¼ z (3.29)
Given the specification above, the condition of efficient pricing implies that the
final commodity price must equal the sum of factor costs and the cost of intermedi-
ate commodities required in the production of a unit of the final commodity. Thus:
A0rwþ A0
qp ¼ p (3.30)
The first term is the price component of rewards to primary factors and the
second term is the price component of rewards to intermediate commodities at their
market prices.49
Solving Eq. 3.30 for p gives:
A0rw ¼ p� A0
qp (3.31)
A0rw ¼ I� A0
q
� �p (3.32)
I� A0q
� ��1A0
rw ¼ p (3.33)
Substituting from Eqs. 3.29 and 3.33 into Eq. 3.26 gives:
Grwþ Gq I� A0q
� ��1A0
rw ¼ Ar½I� Aq��1q (3.34)
Frwþ Fq I� A0q
� ��1A0
r w ¼ q (3.35)
Pre-multiplying Eq. 3.35 by Ar ½I� Aq��1, direct and indirect factor
requirements, gives:
49 This equation is equivalent to the price formulation of input–output analysis. The price system
appears as the dual of the quantity system, and vice versa, and the two can be studied indepen-
dently. Following these principles, we obtain the transpose of Aq and Ar,, which is denoted by A0q
and A0r.
Appendix 1: The Reformulation of the Walras-Cassel Model 45
Ar ½I� Aq��1Frwþ Ar½I� Aq��1Fq½I� A0q��1
A0r w ¼
Ar½I� Aq��1q ð3:36Þ
It follows that:
Gq ¼ Ar½I� Aq��1 Fq (3.37)
Gr ¼ Ar½I� Aq��1 Fr (3.38)
Equations 3.37 and 3.38 specify the effects of commodity demand functions on
factor supplies (direct and indirect factor requirements) necessary for the efficient
production, (3.27) and (3.28), and the efficient pricing condition (3.30) to hold.
Equation 3.37 specifies these conditions on the commodity price matrix assuming
that Fq is specified, and Eq. 3.38 specifies these conditions on the factor price
matrix assuming that Fr is specified. Given the assumptionm ¼ n and the rank of Ar
is equal to n the generalized inverse50 of Ar exists. Thus, applying the generalized
inverse of {Ar [I � Aq]�1} to Eq. 3.38 gives:
Fr ¼ ½I� Aq�½A0r Ar��1A0
rGr (3.39)
Equation 3.39 specifies the generation of the income constraint on demand.
Similarly, Eq. 3.38 specifies the generation of the income constraint on the factor
supply functions. Hence, the commodity demand functions and the factor supply
functions may be specified by the Cassel-Wald specification:
FðpÞ ¼ q and GðwÞ ¼ r (3.40)
which together with Ar and Aq specify a consistent linear system without loss of the
generality of Dorfman, Samuelson and Solow specification of the Walrasian equi-
librium system. As a consequence, commodity prices can be expressed as function
of factor prices alone, using the non-substitution theorem of Samuelson (1951).
The Fr and Gq matrices of the linearised Walras-Cassel model are completely
specified by the Fq, Gr, Ar and Aq matrices together with the conditions of efficient
production, Eqs. 3.27 and 3.28, and the efficient pricing condition (3.30). Thus, the
information contained in Gq and Fr in the Walrasian specification is redundant.
Both functions (F and G) together with the specifications given above specify a
system homogeneous of degree zero in w and p. This implies, that the F and Gfunctions need no longer be specified with homogeneity of degree zero. The
equations in (3.40) can be converted to inverse form:
50 For details, see Penrose, R., (1955). A summary is given in Maddala, G. S., (1977).
46 3 The Planner and the Market: The Takayama Judge Activity Model
w ¼ G�1ðrÞ and p ¼ F�1ðqÞ (3.41)
where G�1 and F�1 are the inverses of G and F, respectively. Hence, the objectionby Dorfman, Samuelson and Solow that this inversion is not admissible in general
does not hold for the linearised Walras-Cassel model.
Appendix 2: Tables 3.2, 3.3, 3.4, 3.5, and 3.6
Table 3.2 Sectors and their definitions in the model
Sector Definition Column Definition
1 Agriculture, fishing 1 Domestic production (Z)
2 Forestry 2 Non-competitive imports (m)
3 Mining and quarrying 3 Competitive imports (M)
4 Sheltered food industry 4 Exports (E), 1980 values
5 Exposed food industry 5 Change in domestic production
6 Beverage and tobacco
industry
6 Change in competitive imports
7 Textile and clothing
industry
7 Change in exports
8 Wood, pulp and paper
industry
8 Capacity utilization in percent of the sectorally
established capital stocks
9 Printing industry 9 Percentage share of domestic production
10 Rubber products
industry
10 Percentage share of competitive imports
11 Chemical industry 11 Net trade ratio (E � M)/(E þ M), 1 only
exports, �1 only imports, 0 balance
12 Petroleum and coal
industry
12 Private consumption (x)
13 Non-metallic mineral
products
13 Equilibrium prices (p) of the quadratic
variables (x) – indexed at 1,000
14 Basic metal industries
15 Engineering, excl.
shipyards
16 Shipyards
17 Other manufacturing
18 Electricity, gas, heating
and water
19 Construction
20 Merchandise trade
21 Transport and
communications
22 Housing
23 Private services
24 Foreign tourist services
Appendix 2: Tables 3.2, 3.3, 3.4, 3.5, and 3.6 47
Table of 1980 statistics – million Skr – 1975 prices
Column
1 2 3 4 5678 9 10 11 12 13Sector
1 14,202 1,863 1,007 1,174 000100 2.69 1.13 0.08 6,617 1,000
2 8,388 284 272 129 000100 1.55 0.30 �0.36 230 1,000
3 4,381 6,371 1,712 2,457 000100 0.81 1.92 0.18 43 1,000
4 23,915 38 1,484 773 000100 4.41 1.66 �0.32 16,549 1,000
5 12,769 645 2,664 758 000100 2.36 2.99 �0.56 9,333 1,000
6 12,149 383 256 93 000100 2.24 0.29 �0.47 11,285 1,000
7 14,439 154 7,636 2,599 000100 2.66 8.54 �0.49 17,549 1,000
8 44,252 51 2,625 19,680 000100 8.17 2.94 0.76 4,374 1,000
9 11,544 0 610 413 000100 2.13 0.68 �0.19 2,772 1,000
10 1,941 19 1,097 599 000100 0.36 1.23 �0.29 991 1,000
11 16,796 995 8,681 6,096 000100 3.10 9.73 �0.18 4,479 1,000
12 19,188 26 6,159 2,300 000100 3.54 6.90 �0.46 6,125 1,000
13 5,878 0 1,447 1,022 000100 1.08 1.62 �0.17 396 1,000
14 18,875 96 5,342 8,123 000100 3.48 5.99 0.21 – –
15 84,100 0 32,90238,045 000100 15.52 36.88 0.07 13,122 1,000
16 5,138 0 722 1,660 000100 0.95 0.81 0.39 1,363 1,000
17 2,908 0 1,045 506 000100 0.54 1.17 �0.35 2,125 1,000
18 11,571 0 110 108 000100 2.14 0.12 �0.01 4,386 1,000
19 49,971 0 0 0 000100 9.22 0 0.00 – –
20 50,818 0 1,230 1,561 000100 9.38 1.38 0.12 – –
21 35,208 0 3,487 7,685 000100 6.50 3.91 0.38 7,047 1,000
22 33,683 0 0 0 000100 6.22 0 0.00 31,459 1,000
23 59,752 0 2,860 3,258 000100 11.03 3.21 0.07 19,719 1,000
24 0 0 5,861 2,960 000– 0 6.57 �0.37 3,171 1,000
Total 541,86610,92589,209101,728000 163,134
48 3 The Planner and the Market: The Takayama Judge Activity Model
Table
3.3
Application1:Tem
porary
equilibrium
–period1
Column
12
34
56
78
910
11
12
13
Sector
115,590
2,082
816
1,174
1,388
�191
0100
2.69
0.91
0.18
7,168
750
29,227
312
266
129
839
�60
100
1.59
0.20
0.35
238
750
34,819
7,008
1,271
2,457
438
�441
0100
0.83
1.42
0.32
68
750
426,306
42
2,642
773
2,391
1,158
0100
4.55
2.94
�0.55
19,453
750
514,044
709
2,549
758
1,275
�115
0100
2.43
2.84
�0.54
10,005
750
613,364
421
093
1,215
�256
0100
2.31
01.00
12,150
735
715,884
169
11,219
2,599
1,445
3,583
0100
2.75
12.50
�0.62
22,504
750
848,675
56
256
19,680
4,423
�2,369
0100
8.41
0.29
0.97
5,027
750
912,699
0465
413
1,155
�145
0100
2.19
0.52
�0.06
3,219
750
10
2,135
21
1,117
599
194
20
0100
0.37
1.24
�0.30
1,101
750
11
18,476
1,094
8,659
6,095
1,680
�22
0100
3.19
9.65
�0.17
5,222
750
12
21,113
29
5,789
2,300
1,925
�370
0100
3.65
6.45
�0.43
6,677
750
13
6,466
01,082
1,022
588
�365
0100
1.12
1.21
�0.03
456
750
14
16,422
84
7,426
8,123
�2,453
2,084
079
2.84
8.28
0.04
––
15
92,508
027,832
38,045
8,408
�5,070
0100
15.99
31.02
0.16
14,683
750
16
00
5,917
1,660
�5,138
5,195
00
06.59
�0.56
1,637
750
17
3,200
0967
506
292
�78
0100
0.55
1.08
�0.31
2,287
750
18
12,399
00
108
828
�110
097
2.14
01.00
4,744
226
19
50,637
00
0666
00
92
8.75
00.00
––
20
55,899
061
1,561
5,081
�1,169
0100
9.66
0.07
0.92
––
21
38,728
02,716
7,685
3,520
�771
0100
6.69
3.03
0.48
7,944
750
22
34,298
00
0615
00
93
5.92
00.00
32,074
181
23
65,729
01,687
3,258
5,977
�1,173
0100
11.36
1.88
0.32
21,460
750
24
00
6,991
2,690
01,130
0–
07.79
�0.44
4,301
750
Total
578,617
12,027
89,728
101,728
36,751
519
0182,418
Appendix 2: Tables 3.2, 3.3, 3.4, 3.5, and 3.6 49
Table
3.4
Application2:Tem
porary
equilibrium
–period2
Column
12
34
56
78
910
11
12
13
Sector
117,149
2,290
01,174
1,559
�816
0100
2.90
01.00
7,135
1,015
29,975
338
0129
748
�266
098
1.69
01.00
260
436
35,162
7,507
02,457
343
�1,271
097
0.87
01.00
64
1,042
428,937
46
199
773
2,631
�2,443
0100
4.89
0.22
0.59
18,933
1,045
515,448
780
1,557
758
1,404
�992
0100
2.61
1.75
�0.35
9,884
1,045
614,700
463
093
1,336
00
100
2.48
01.00
13,383
624
717,474
186
8,947
2,599
1,590
�2,272
0100
2.95
10.07
�0.55
21,616
1,045
849,430
57
019,680
755
�256
092
8.35
01.00
5,925
655
913,969
00
413
1,270
�465
0100
2.36
01.00
3,591
792
10
2,349
23
983
599
214
�134
0100
0.40
1.11
�0.24
1,081
1,045
11
20,323
1,204
7,383
6,095
1,847
�1,276
0100
3.43
8.31
�0.10
5,089
1,045
12
23,227
31
3,175
2,300
2,114
�2,614
0100
3.93
3.57
�0.16
6,578
1,045
13
6,933
00
1,022
467
�1,082
097
1.17
01.00
527
703
14
202
120,093
8,123
�16,220
12,667
01
0.03
22.62
�0.42
––
15
101,760
032,424
38,045
9,252
4,592
0100
17.20
36.50
0.08
14,403
1,045
16
00
6,855
1,660
0938
0-
07.72
�0.61
1,588
1,045
17
3,519
0427
506
319
�540
0100
0.59
0.48
0.08
2,258
1,045
18
12,429
00
108
30
00
91
2.10
01.00
5,057
324
19
41,088
00
0�9
,549
00
74
6.94
00.00
––
20
57,836
00
1,561
1,937
�61
094
9.77
01.00
––
21
42,602
00
7,685
3,874
�2,716
0100
7.20
01.00
8,023
978
22
34,870
00
0572
00
92
5.89
00.00
32,646
238
23
72,302
00
3,258
6,573
�1,687
0100
12.22
01.00
22,248
887
24
00
6,789
2,690
0�2
02
0–
07.64
�0.43
4,099
1,045
Total
591,683
12,926
88,832
101,728
13,066
�896
0184,388
50 3 The Planner and the Market: The Takayama Judge Activity Model
Table
3.5
Application3:Tem
porary
equilibrium
–period3
Column
12
34
56
78
910
11
12
13
Sector
17,912
1,057
11,689
1,174
�9,237
11,689
042
1.30
12.28
�0.82
7,569
803
29,638
326
0129
�337
00
88
1.58
01.00
283
328
31,464
2,129
9,622
2,457
�3,698
9,622
026
0.24
10.11
�0.59
84
803
431,831
51
259
773
2,894
60
0100
5.21
0.27
0.50
21,220
803
516,992
858
0758
1,544
�1,557
0100
2.78
01.00
10,774
668
616,168
510
093
1,468
00
100
2.65
01.00
14,789
570
719,218
205
11,101
2,599
1,744
2,154
0100
3.15
11.67
�0.62
25,541
803
851,131
59
019,680
1,701
00
94
8.38
01.00
7,165
524
915,236
00
413
1,267
00
99
2.50
01.00
4,407
544
10
2,584
25
916
599
235
�67
0100
0.42
0.96
�0.21
1,168
803
11
22,359
1,324
6,211
6,095
2,033
�1,172
0100
3.66
6.53
�0.01
5,674
803
12
25,549
35
1,372
2,300
2,322
�1,803
0100
4.19
1.44
0.25
7,012
803
13
6,808
00
1,022
�125
00
87
1.12
01.00
636
547
14
00
21,104
8,123
�202
1,011
00
022.18
�0.44
––
15
111,940
017,665
38,045
10,180
�14,759
0100
18.34
18.56
0.37
15,631
803
16
00
7,296
1,660
0441
0-
07.67
�0.63
1,804
803
17
3,872
0250
506
353
�177
0100
0.63
0.26
0.34
2,385
803
18
12,842
00
108
413
00
94
2.10
01.00
5,404
250
19
37,409
00
0�3
,679
00
61
6.13
00.00
––
20
60,381
00
1,561
2,545
00
89
9.89
01.00
––
21
44,496
00
7,685
1,894
00
95
7.29
01.00
9,930
468
22
35,493
00
0623
00
86
5.81
00.00
33,269
170
23
77,067
00
3,258
4,765
00
97
12.63
01.00
26,187
434
24
00
7,679
2,690
0890
0–
08.07
�0.48
4,989
803
Total
610,387
6,578
95,164
101,728
18,704
6,332
0205,921
Appendix 2: Tables 3.2, 3.3, 3.4, 3.5, and 3.6 51
Table
3.6
Application4:Tem
porary
equilibrium
–period4
Column
12
34
56
78
910
11
12
13
Sector
121
321,893
1,174
�7,891
10,204
00.24
022.32
�0.90
8,446
602
29,943
337
0129
305
00
94
1.59
01.00
308
240
30
013,973
2,457
�1,464
4,351
00
014.25
�0.70
124
602
435,014
56
2,372
773
3,183
2,113
0100
5.60
2.42
�0.51
25,839
602
517,587
888
0758
595
00
94
2.81
01.00
12,200
469
617,787
561
093
1,619
00
100
2.84
01.00
16,348
524
721,140
225
17,057
2,599
1,922
5,956
0100
3.38
17.39
�0.74
33,451
602
852,694
61
019,680
1,563
00
94
8.43
01.00
8,748
393
916,752
00
413
1,516
00
100
2.68
01.00
5,457
413
10
2,842
28
908
599
258
�80
100
0.45
0.93
�0.21
1,343
602
11
24,591
1,457
5,474
6,095
2,235
�737
0100
3.93
5.58
0.05
6,856
602
12
28,102
38
02,300
2,553
�1,372
0100
4.49
01.00
8,114
500
13
6,318
00
1,022
�490
00
84
1.01
01.00
778
408
14
00
21,506
8,123
0402
0–
021.93
�0.45
––
15
118,694
00
38,045
6,754
�17,665
096
18.98
01.00
19,311
409
16
00
5,266
1,660
0�2
,030
0–
05.37
�0.52
2,239
602
17
4,258
0145
506
386
�105
0100
0.68
0.15
0.55
2,642
602
18
13,368
00
108
526
00
95
2.14
01.00
5,789
168
19
28,456
00
0�8
,953
00
69
4.55
00.00
––
20
62,939
00
1,561
2,558
00
95
10.07
01.00
––
21
47,320
00
7,685
2,824
00
97
7.57
01.00
12,285
343
22
36,152
00
0659
00
93
5.78
00.00
33,928
122
23
81,343
00
3,258
4,276
00
96
13.01
01.00
30,911
321
24
00
9,476
2,690
01,797
0–
09.66
�0.56
6,786
602
Total
625,322
3,653
98,070
101,728
14,935
2,906
0241,903
52 3 The Planner and the Market: The Takayama Judge Activity Model
References
Armington P (1969) A theory of demand for products distinguished by place of production. IMF
Staff Pap 16:159–178
Bergman L (1986) ELIAS: a model of multisectoral economic growth in a small open economy.
IIASA, Laxenburg
Blattner N (1977) Intraindustrieller Aussenhandel; Empirische Beobachtungen im Falle der
Schweiz und Teoretische Interpretationen. Weltwirdschaftliches Archiv 113(1):88–103
Dervis K, de Melo J, Robinson S (1982) General equilibrium models for development policy.
Cambridge University Press, Cambridge
Dorfman R, Samuelson PA, Solow RM (1958) Linear programming and economic analysis.
McGraw-Hill, New York, pp 346–389
Ems E (1988) Exportindustrins framtid, SIND 1988:2. Stockholm
Enke S (1951) Equilibrium among spatially separated markets: solutions by electric analogue.
Econometrica 19:40–47
Erixon L (1988) Loner och konkurrenskraft – lonekostnadernas betydelse for Sveriges
marknadsandelar. I Ems (ed) Exportindustrins Framtid, SIND 1988:2
Flam H (1981) Growth, allocation and trade in Sweden, vol 12, Institute for International
Economic Studies, Monograph series. University of Stockholm, Stockholm
Grubel HG, Lloyd PJ (1975) Intra-Industry trade: the theory and measurement of international
trade in differentiated products. Macmillan/Halsted, London
Harrington DH (1973) Quadratic input–output analysis: methodology for empirical general equi-
librium models. Ph.D. dissertation, Purdue University, West Lafayette
Hine RC (1988) 1992 and the pattern of specialisation in European industry. Paper presented to a
workshop on the Nordic countries an the EEC, Lund
Hotelling H (1932) Edgeworth’s taxation paradox and the nature of demand and supply functions.
J Polit Econ 40:557–616
Kirzner IM (1973) Competition and entrepreneurship. The University of Chicago Press, Chicago
Koopmans TC (1957) Three essays on the state of economic science. McGraw-Hill, New York
Krugman P (1980) Scale economies, product differentiation and the pattern of trade. Am Econ Rev
70(5):950–959
Lancaster KJ (1980) Intra-industry trade under perfect monopolistic competition. J Int Econ
10:151–175
Lundberg L (1988) The Nordic countries and economic integration in Europe: trade barriers and
patterns of trade and specialization. Trade Union Institute of Economic Research, Stockholm
Maddala GS (1977) Econometrics. McGraw-Hill, New York
Marshall A (1925) Principles of economics, 8th edn. Macmillan, London
McCarl BA, Spreen TH (1980) Price endogenous mathematical programming as a tool for sector
analysis. Am J Agric Econ 62:87–102
Noren R (1987) Comparative advantages revealed. Experiments with a quadratic programming
model of Sweden. Ph.D. dissertation. Department of Economics, University of Stockholm.
Akademitryck, Edsbruk
Noren R (1991) A Takayama-judge activity model applied to the analysis of allocation and trade in
Sweden. Appl Econ 23:1201–1212
Norman VD (1983) En Liten, Apen Okonomi. Universitetsforlaget, Oslo
Penrose R (1955) Generalized inverse for matrices. In: Proceedings of the Cambridge philosophi-
cal society, Cambridge, pp 406–413
Petersson L (1984) Svensk Utrikeshandel 1871–1980. En studie I den den intraindustriella
handelns framvaxt. Lund Economic Studies, Lund
Plessner Y (1965) Quadratic programming competitive equilibrium models for the U.S agricul-
tural sector. Ph.D. dissertation. Iowa State University, Ames
Plessner Y (1967) Activity analysis, quadratic programming and general equilibrium. Int Econ
Rev 8:168–179
References 53
Plessner Y, Heady E (1965) Competitive equilibrium solutions with quadratic programming.
Metroeconomica 17:117–130
Samuelson P (1951) Abstract of a theorem concerning substitutability in open Leontief models. In:
Koopmans TC (ed) Activity analysis of production and allocation. Wiley, New York
Samuelson PA (1952) Spatial price equilibrium and linear programming. Am Econ Rev
42:283–303
Swedenborg B (1988) The EC and the locational choice of Swedish multinational companies.
Paper presented to a workshop on the Nordic countries an the EEC, Lund
Sodersten B (1980) International economics. Macmillan, London
Takayama T, Judge G (1964a) Equilibrium among spatially separated markets: a reformulation.
Econometrica 32:510–524
Takayama T, Judge G (1964b) An interregional activity analysis model of the agricultural sector.
J Farm Econ 46:349–365
Takayama T, Judge G (1971) Spatial and temporal price and allocation models. North-Holland,
Amsterdam
Taylor L (1975) Theoretical foundations and technical implications. In: Blitzer CR et al (eds)
Economy-wide models and development planning. Oxford University Press, Oxford
Varian RH (1984) Microeconomic analysis, 2nd edn. W.W. Norton, New York
Wald A (1951) On some systems of equations of mathematical economics. Econometrica
19:368–403
Werin L (1965) A study of production, trade and allocation of resources, vol 6, Stockholm
economic studies, new series. Almqvist & Wiksell, Stockholm
Yaron D, Plessner Y, Heady E (1965) Competitive equilibrium application of mathematical
programming. Can J Agr Econ 13:65–79
Zusman P (1969) The stability of interregional competition and the programming approach to the
analysis of spatial trade equilibria. Metroeconomica 21:55
Government Publications
SOU 1984:7. LU 84 (The medium term survey of the Swedish economy), Appendix 17.
SM 1974:52. Stocks of fixed capital 1950–73 and capital consumption 1963–73, Swedish Central
Bureau of Statistics
Statistical Reports SM N 1970:13. Input–output tables for Sweden 1964, Swedish Central Bureau
of Statistics
Statistical Reports. SM N 1981:2.5. Appendix 1, Final consumption expenditure, Swedish CentralBureau of Statistics
Statistical Reports. SM N 1981:2.5. Appendix 2, Capital formation and stocks of fixed capital,Swedish Central Bureau of Statistics
Statistical Reports. SM N 1981:2.5. Appendix 5, Employment and compensation of employments,
Swedish Central Bureau of Statistics
54 3 The Planner and the Market: The Takayama Judge Activity Model