Barry T.

Coyle

Production Economics
This is a draft: please do not distribute
c Copyright; Barry T. Coyle, 2010

October 9, 2010

University of Manitoba
Foreword

This booklet is typed based on professor Barry Coly’s lecture notes for ABIZ 7940
Production Economics in Winter 2008. I am responsible for all the errors and typos.

Winnipeg, October 2010 Ning Ma

v
Contents

1 Static Cost Minimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Properties of c.w; y/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Corresponding properties of x.w; y/ solving problem (1.1) . . . . . . . 3
1.3 Second order relations between c.w; y/ and f .x/ . . . . . . . . . . . . . . . 4
1.4 Additional properties of c.w; y/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.5 Applications of dual cost function in econometrics . . . . . . . . . . . . . . . 7
1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2 Static Competitive Profit Maximization . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1 Properties of .w; p/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Corresponding properties of y.w; p/ and x.w; p/ . . . . . . . . . . . . . . . 13
2.3 Second order relations between .w; p/ and f .x/, c.w; y/ . . . . . . . 14
2.4 Additional properties of .w; p/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5 Le Chatelier principles and restricted profit functions . . . . . . . . . . . . . 17
2.6 Application of dual profit functions in econometrics: I . . . . . . . . . . . . 19
2.7 Industry profit functions and entry and exit of firms . . . . . . . . . . . . . . 21
2.8 Application of dual profit functions in econometrics: II . . . . . . . . . . . 23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3 Static Utility Maximization and Expenditure Constraints . . . . . . . . . . . 27

3.1 Properties of V .p; y/ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Corresponding properties of x.p; y/ solving problem (3.1) . . . . . . . . 31
3.3 Application of dual indirect utility functions in econometrics . . . . . . 32
3.4 Profit maximization subject to budget constraints . . . . . . . . . . . . . . . . 34
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4 Nonlinear Static Duality Theory (for a single agent) . . . . . . . . . . . . . . . . 39

4.1 The primal-dual characterization of optimizing behavior . . . . . . . . . . 39
4.2 Producer behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Consumer behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

vii
viii Contents

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5 Functional Forms for Static Optimizing Models . . . . . . . . . . . . . . . . . . . 47

5.1 Difficulties with simple linear and log-linear Models . . . . . . . . . . . . . 47
5.2 Second order flexible functional forms . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.3 Examples of second order flexible functional forms . . . . . . . . . . . . . . 51
5.4 Almost ideal demand system (AIDS) . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.5 Functional forms for short-run cost functions . . . . . . . . . . . . . . . . . . . 58
5.5.1 Normalized quadratic: c  D c=w0 , w  D w=w0 . . . . . . . . . . 59
5.5.2 Generalized Leontief: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.5.3 Translog: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

6 Aggregation Across Agents in Static Models . . . . . . . . . . . . . . . . . . . . . . . 61

6.1 General properties of market demand functions . . . . . . . . . . . . . . . . . 61
6.2 Condition for exact linear aggregation over agents . . . . . . . . . . . . . . . 64
6.3 Linear aggregation over agents using restrictions on the
distribution of output or expenditure . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.4 Condition for exact nonlinear aggregation over agents . . . . . . . . . . . . 75
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

7 Aggregation Across Commodities: Non-index Number Approaches . . 77

7.1 Composite commodity theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
7.2 Homothetic weak separability and two-stage budgeting . . . . . . . . . . . 79
7.3 Implicit separability and two-stage budgeting . . . . . . . . . . . . . . . . . . . 83
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

8 Index Numbers and Flexible Functional Forms . . . . . . . . . . . . . . . . . . . . 87

8.1 Laspeyres index numbers and linear functional forms . . . . . . . . . . . . 87
8.2 Exact indexes for Translog functional forms . . . . . . . . . . . . . . . . . . . . 89
8.3 Exact indexes for Generalized Leontief functional forms . . . . . . . . . . 96
8.4 Two-stage aggregation with superlative index numbers . . . . . . . . . . . 97
8.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
8.6 Laspeyres and Paasche cost of living indexes . . . . . . . . . . . . . . . . . . . 102
8.7 Fisher indexes (for inputs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

9 Measuring Technical Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

9.1 Dual cost and profit functions with technical change . . . . . . . . . . . . . 106
9.2 Non-Parametric index number calculations of changes in
productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
9.3 Parametric index number calculations of changes in productivity . . . 113
9.4 Incorporating variable utilization rates for capital . . . . . . . . . . . . . . . . 117
9.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Chapter 1
Static Cost Minimization

Consider a firm producing a single output y using N inputs x D .x1 ;


; xN /

according to a production function y D f .x/. The firm is a price taker in its N
factor markets, i.e. the firm treats factor prices w D .w1 ;


; wN / as given. We
assume that all inputs are freely adjustable and perfect rental markets exist for all
capital goods, i.e. for now we ignore all costs of adjustment that can lead to dynamic
behavior.
A necessary condition for static profit maximization is static cost minimization,
i.e. at the profit maximizing level of output y and input prices w the firm necessarily
solves the following cost minimization problem:
N
X
min wi xi D wx 
x0
i D1
(1.1)
s.t. f .x/  y
The minimum cost c  wx  to problem (1.1) depends on the levels of input
prices w and output y, and of course on the production function y D f .x/. By
solving (1.1) using different value of .w; y/ we can in principle trace out the relation
between minimum cost c and parameters .w; y/, conditional on the firm’s particular
production function y D f .x/. This relation c D c.w; y/ between minimum cost
and parameters .w; y/ is called the firm’s dual cost function.

1.1 Properties of c.w; y/

Property 1.1.
a) c.w; y/ is increasing (or, more precisely, non-decreasing) in all parameters
.w; y/.

1
2 1 Static Cost Minimization

Fig. 1.1 c.w; y/ is concave c(wA , wB , y)

in w

wA

b) c.w; y/ is linear homogeneous in w. i.e. c.w; y/ D c.w; y/ for all scaler

 > 0 (if all factor prices w increase by the same proportion. e.g. 10%, then
the minimum cost of attaining the same level of output y also increases by
this proportion)
c) c.w; y/ is concave in w. i.e. c.wA C .1 /wB ; y/  c.wA ; y/ C .1
/c.wB ; y/ for all 0    1. See Figure 1.1
d) If c.w; y/ is differentiable in w, then

@c.w; y/
xi .w; y/ D
@wi

(Shephard’s Lemma)1

Proof. Property 1.1.a is obvious from Equation (1.1), and 1.1.b follows from the
fact that an equiproportional change in all factor prices w does not change relative
factor prices and hence does not change the cost-minimizing level of inputs x  for
problem (1.1). 1.1.c is not so obvious. In order to prove it simply note that, for
wC  wA C .1 /wB and xC solving (1.1) for .wC ; y/,

c.wC ; y/  wC xC
D wA xC C .1 /wB xC (1.2)
 c.wA ; y/ C .1 /c.wB ; y/

since wA xC  c.wA ; y/ and wB xC  c.wB ; y/ (e.g. wA xC cannot be less than
the minimum cost for problem (1.1) given prices wA ¤ wC : wA xC > c.wA ; y/
unless xC solves (1.1) for prices wA as well as for prices wC ). Numerous proofs of
Shephard’s Lemma 1.1.d are available. Here we simply consider the most obvious
method of proof (see Varian 1992 for alternative methods).
Expressing (1.1) in Lagrange form
1
Note that c.w; y/ can be differentiable in w even if, e.g. the production function y D f .x/ is
Leontief (fixed proportions). In general differentiability of c.w; y/ is a weaker assumption than
differntiability of y.
1.2 Corresponding properties of x.w; y/ solving problem (1.1) 3

Q .f .x/ y/ D wx  Q  f .x  /


min wx y D c.w; y/ (1.1’)
Q
x;

with first order condition

@f .x  /
wi Q  D0
@xi
f .x  / y D 0
@c.w;y/
Then @wi
can be calculated by total differentiation as follows:

N 
X @xj @f .x  / @Q 
 
@c.w; y/
D xi C Q  f .x  /


wj y
@wi @wi @xj @wi (1.3)
j D1

D xi

by the first order conditions to problem (1.1’). u

t
It is important to note that Shephard’s Lemma 1.1.d is simply an application of
the envelope theorem (Samuelson 1947). The lemma states that, for an infinitesimal
change in factor price wi (all other factor prices and output remaining constant), the
change in minimum cost divided by the change in wi is equal to the equilibrium
level of input i in the absence of any change in .w; y/. In other words, in the limit,
zero changes in equilibrium x  in response to a change in wi are optimal. Obviously
such a lemma has no economic content, i.e. does not describe optimal response to
finite changes in wi . Nevertheless Shephard’s and analogous envelope theorems are
critical to the empirical and theoretical application of duality theory. This distinction
is easily missed in more complex models.

1.2 Corresponding properties of x.w; y/ solving problem (1.1)

Property 1.2.
a) x.w; y/ is homogeneous of degree 0 in w. i.e. x.w; y/ D x.w; y/ for all
scalar
h  i 0.
@x.w;y/
b) @w
is symmetric negative semidefinite.
N N

Proof. 1.2.a simply states that the cost minimizing solution x  to problem (1.1) de-
pends
h 2 only i on relative prices. In order to prove 1.2.b, note that the Hessian matrix
@ c.w;y/
@w@w
is symmetric negative semidefinite by concavity and twice differen-
N N
@c.w;y/
tiability of c.w; y/ in w, and then note that @w
D x.w; y/ for all w (Shephard’s
h 2 i h i
Lemma) ) @ @w@w c.w;y/
D @x.w;y/
@w
. t
u
N N N N
4 1 Static Cost Minimization

In order to test economics theories it is important to know all of the restrictions

that are placed on observable behavior by particular theories. This is known as the
integrability problem in economics. It can easily be shown that 1.2.a-b exhausted
the (local) properties that are placed on factor demands x.w; y/ by the hypothesis
of cost minimization (1.1).

Proof. It has already been shown that the properties 1.1 of the cost function imply
1.2. In order to show that 1.2 exhausts the the implications of cost minimization
(1.1) for local properties of x.w; y/, first total differentiate c.w; y/  wx.w; y/
with respect to wi ,
N
@c.w; y/ X @xj .w; y/
 xi .w; y/ C wj i D 1;


; N: (1.4)
@wi @wi
j D1

PN @xj .w;y/
1.2.a implies (by Euler’s theorem2 ) j D1 wj @wi
D 0 (i D 1;


; N ) and to-
gether with symmetry 1.2.b this reduces the identity (1.4) to @c.w;y/
@w
D xi .w; y/ 
h i i
@x.w;y/
0, (i D 1;


; N ) (Shephard’s Lemma). In addition @w
1.2.b implies
N N
that the system of differential equations xi .w; y/ D @c.w;y/
@wi
.i D 1;


; N / inte-
grates up to an underlying cost function c.w; y/ (Frobenius theorem3 ). Shephard’s
Lemma also implies
h (by i of x.w; y/ D @c.w; y/=@w with
i simple differentiation
h 2
@x.w;y/ @ c.w;y/
respect to w) @w
D which is negative semidefinite
N N h@w@w
2
NiN
by 1.2.b. It can then be shown that @ @w@wc.w;y/
negative semidefinite implies
N N
y D f .x/ is quasiconcave at x D x.w; y/ (see (1.6) below). This establishes
the second order conditions on the production function y D f .x/ for competitive
cost minimization. The first order condition follow from the fact that 1.2 establishes
Shephard’s Lemma for all @x.w;y/
@w
, @.w;y/
@w
, satisfying 1.2 (see (1.3)). u
t

1.3 Second order relations between c.w; y/ and f .x/

It is sometimes interesting to ask whether or not the firm’s production function

y D f .x/ can be recovered from knowledge of the firm’s cost function c.w; y/,
i.e. can we construct y D f .x/ directly from knowledge of c.w; y/? The answer is
essentially yes (the only qualification is that we cannot recover f .x/ at levels of x
that cannot be solutions to a cost minimization problem (1.1) for some .w; y/, i.e. at
levels of associated with locally non-convex isoquants). For example, given knowl-

2
Euler’s theorem states that ,if g.x/ D r g.x/ for all scaler  > 0 (i.e. the function g.x/ is
@2 g.x/
homogenous of degree r), then rg.x/ D i xi @g.x/ and .r 1/ @g.x/
P P
@xi @xj
D i xi @x i @xj
3 @ .v/
The Frobenius theorem states that a system of differential equations gi .v/ D @vi
.i D
@gi .v/ @gj .v/
1;


; T / has a solution .v/ if and only if @vj
D @vi
.i; j D 1;


; T /.
1.3 Second order relations between c.w; y/ and f .x/ 5

edge of c.w; y/ and differentiability of c.w; y/, application of Shephard’s Lemma

@c @c
@w1
D x1 ;


; @wN
D xN immediately gives the cost-minimizing levels of inputs
x corresponding to .w; y/ (this assumes a unique solution x to problem (1.1)). By
varying .w; y/ we can map out y D f .x/ from c.w; y/ in this manner.
A related point that will be important later (in a lecture on functional forms) is
that the first and second derivatives of f .x/ at x.w; y/ can be calculated directly
from knowledge of the first and second derivatives of c.w; y/. The first derivatives
can be calculated simply as

@f .x.w; y// wi
D i D 1;


; N (1.5)
@xi @c.w; y/=@y

using the first order conditions @f@x.x/ Q i D 0 .i D 1;


; N / for cost minimiza-

w
i
tion (1.1’), where Q  @c.w;y/
@y
. The procedure for calculating the second derivatives
of f .x/ from c.w; y/ is not quite as obvious. The corresponding formula in matrix
notation is
" @2 c.w;y/ @2 c.w;y/ # 1
@2 f
" #
@c.w;y/ @f
˝
@y
@f
@x@x @x D @2@w@w @w@y
c.w;y/ @2 c.w;y/
(1.6)
@x
0 .N C1/.N C1/ @y@w @y@y .N C1/.N C1/

where it is assumed without loss of generality that the above inverse exists4 .

Proof. Consider the case N D 2 so y D f .x1 ; x2 / and c D c.w1 ; w2 ; y/. The first
order condition for cost minimization (1.1’) can be written as

w1 cy .w; y/fx1 D 0
w2 cy .w; y/fx2 D 0 (1.7)
y D f .x/

where for now cy .w; y/  @c.w;y/

@y
, fx1  @f@x.x/
1
, etc.
Total differentiating these conditions (1.7) with respect to .w1 ; w2 ; y/ yields (us-
ing Shephard’s Lemma):

PN 2
4
c.w; y/ linear homogenous in w and Euler’s theorem imply 0 D j D1 wj @@wc.w;y/ i @wj
.i D
h 2 i
1;


; N /, which implies @ @w@w
c.w;y/
does not have full rank. Nevertheless the above bor-
" 2 2
# N N
@ c.w;y/ @ c.w;y/
@w@w @w@y
dered matrix @2 c.w;y/ @2 c.w;y/
generally have full rank.
@y@w @y@y .N C1/.N C1/
6 1 Static Cost Minimization
8
<1
cyw1 fx1 cy fx1 x1 cw1 w1 cy fx1 x2 cw1 w2
ˆ D0
@
) cyw1 fx2 cy fx1 x2 cw1 w1 cy fx2 x2 cw1 w2 D0
@w1 ˆ
fx1 cw1 w1 fx2 cw1 w2 D0
:
8
ˆ cyw2 fx1 cy fx1 x1 cw1 w2 cy fx1 x2 cw2 w2 D0
@ <
) 1 cyw2 fx2 cy fx1 x2 cw1 w1 cy fx2 x2 cw2 w2 D0 (1.8)
@w2 ˆ
fx1 cw1 w2 C fx2 cw2 w2 D0
:
8
@ < cyy fx1 cy fx1 x1 cw1 y cy fx1 x2 cw2 y D 0
ˆ
) cyy fx2 cy fx1 x2 cw1 y cy fx2 x2 cw2 y D 0
@y ˆ
fx1 cw1 y C fx2 cw2 y D 0
:

Writing (1.8) in matrix notation,

2 32 3 2 3
cw1 w1 cw1 w2 cyw1 cy fx1 x1 cy fx1 x2 fx1 1 0 0
4 cw1 w2 cw2 w2 cyw2 5 4 cy fx1 x2 cy fx2 x2 fx2 5 D 4 0 1 05 (1.9)
cyw1 cyw2 cyy fx1 fx2 0 0 0 1

t
u

1.4 Additional properties of c.w; y/

Property 1.3.
a) y D f .x/ homothetic , c.w; y/ D .y/ c.w; 1/ for some function .
b) y D f .x/ constant returns to scale , c.w; y/ D y c.w; 1/.
c) All the partial elasticity of substitution between inputs i and j (output y
constant)
@.xi .w;y/=xj .w;y//
xi =xj
ij .w; y/ 
@.wi =wj /
wi =wj

can be calculated simply as

2
c.w; y/ @@wc.w;y/
i @wj
ij .w; y/ D @c.w;y/ @c.w;y/
@wi @wj

(see Uzawa 1962, p. 291-9).

d) Assuming a vector of outputs y D y1 ;


; yM . The transformation function
f .x; y/ D 0 is disjoint (i.e. y1 D f1 .x1 /;


; yM D fM .xM / where input
2
vector x1 ;


; xM do not overlap) only if @@yc.w;y/
i @yj
D 0 for all i ¤ j , all
.w; y/.
1.5 Applications of dual cost function in econometrics 7

1.5 Applications of dual cost function in econometrics

The above theory is usually applied by first specifying a functional form .w; y/
for the cost function c.w; y/ and differentiating .w; y/ with respect to w in order
to obtain the estimating equations

@.w; y/
xi D i D 1;


; N (1.10)
@wi
2 2
(employing Shephard’s Lemma). Then the symmetry restrictions @w@i @w

D @w@j @w

h 2 i j i
@ 
.i D 1;


; N / are tested and the second order condition @w@w negative
N N
semidefinite is checked at all data points .w; y/.
For example a cost function could be postulated as having the functional form
N X
N
X 1 1
cDy aij wi2 wj2
iD1 i D1

(a Generalized Leontief functional form with y D f .x/ showing constant returns

to scale), which leads to the following equations for estimation:
N   21
xi X wj
D aij i D 1;


; N (1.11)
y wi
j D1

@xi @x
Here the symmetry restriction @w j
D @wji are expressed as aij D aj i .i D
1;


; N / which are easily tested. Equation (1.10) can be interpreted as being de-
rived from a cost function c.w; y/ for a producer showing static, competitive cost-
minimizing behavior if and only if the symmetry and second order conditions are
satisfied.
The major advantage of this approach is that it permits the specification of a
system of factor demand equations x D x.w; y/ that are consistent with cost min-
imization and with a very general specification of technology. In contrast, suppose
that we wished to specify explicitly a solution to a cost minimization problem. Then
we would estimate a production function directly with first order conditions for cost
minimization:
8 1 Static Cost Minimization

y D f .x/
@f =@xi wi
D
@f =@xj wj
(1.12)
@f =@xi wi
D
@f =@xk wk

However, unless a very restrictive functional form is specified for the production
function (e.g. Cobb-Douglas), then we can seldom drive the factor demand equa-
tions x D x.w; y/ explicitly from (1.12). Since policy makers are usually more
interested in demand and supply behavior than in production functions per se, this
greater ease in specification of x.w; y/ is an important advantage of duality theory.
Two other advantages of a duality approach rather than a primal approach (1.12)
to the estimation of producer behavior are apparent. First, the hypothesis of compet-
itive cost minimization is more readily tested in the framework of equations (1.10)
than (1.12). Second, variables that are omitted from the econometric model (but
are observed by producers) influence both the error terms and production decisions
but do not necessarily influence factor prices to the same degree (e.g. factor prices
may be exogenous to the industry). This tends to introduce greater simultaneous
equations biases into the estimation of (1.12) than of (1.10).
One disadvantage of (1.10) is that output y, as well as factor prices w, is treated
as exogenous. Since the firm generally in effect chooses y jointly with x , this mis-
specification can lead to simultaneous equations biases in the estimators. So the
extent that production is constant return to scale with a single output, this difficulty
can be avoided by using 1.3.b to specify a unit cost function c.w/ D c.w; y/=y and
applying Shephard’s Lemma to obtain estimating equations

xi @c.w/
D i D 1;


; N (1.13)
y @wi

for a given functional form c.w/ (e.g. (1.11)). Under constant returns to scale
xi .w; y/=y depends on w but not on y, so that (1.13) is well defined and estimation
is independent of whether y is endogenous or exogenous to the firm.
A second disadvantage of this duality approach to the specification of functional
forms for econometric models, and a disadvantage of primal approaches such as
(1.12) as well, is that it is derived from the theory of the individual firm but is
usually applied to market data that is aggregated over firms. Difficulties raised by
such aggregation will be discussed in a later lecture.

1.6 Conclusion

The dual cost function approach offers many advantages in the estimation of pro-
duction technologies. The estimated factor demands x D x.w; y/ measure factor
References 9

substitution along an isoquant and the effects of scale of output on factor demands,
and first and second derivatives of the production function can be calculated. More-
over the assumption of cost minimization is consistent with various broader theories
of producer behavior.
However effective policy making depends more on a knowledge of producer be-
havior than of production functions per se. By ignoring the effect of output prices
on the firm’s input levels, the dual cost function approach generally is inappropriate
for the modeling of economic behavior.
Of course cost functions can be embedded within a broader behavioral model.
For example, static competitive profit maximization implies a cost minimization
model such as 1.1 together with first order conditions

@c.w; y/
Dp (1.14)
@y
for optimal output levels (marginal cost equals output price). This equation implic-
itly defines the optimal level of y as y D y.w; p/ provided of course that y en-
ters (1.14), i.e. @c.w;y/
@y
is not independent of y or equivalently f .x/ does not show
constant returns to scale. In this case equations 1.1.d, (1.10) and (1.14) can be esti-
mated jointly, and the second order condition for profit maximization are expressed
2 c.w;y/
as c.w; y/ concave and @ @y@y  0.
Nevertheless there can be substantial disadvantages to this approach to model-
ing competitive profit maximization. The assumption of constant returns to scale is
commonly employed in empirical studies, and the assumption of profit maximiza-
tion is not so easily tested here (the homogeneity and reciprocity condition for cost
minimization do not imply integration up to a profit function). Therefore, for policy
purposes, it is often better to model and test directly (using dual profit functions) the
hypothesis of competitive profit maximization behavior.5

References

1. Samuelson, Paul A., 1947, Enlarged ed., 1983. Foundations of Economic Analysis, Harvard
University Press. ISBN 0-674-31301-1
2. Uzawa, H. (October 1962). Production Function with Constant Elasticities of Substitution,
Review of Economics Studies, Vol. 29, pp. 291-299.
3. Varian, H. R. (1992). Microeconomic Analysis, Third Edition. W. W. Norton & Company, 3rd
edition.

5
In passing, note that this indirect approach to the modeling of profit maximization (1.1.d, (1.10),
(1.14)) may be superior to a direct approach (see next lecture) when there is substantially higher
multicollinearity between factor prices w and output prices p than between w and output levels
y.
Chapter 2
Static Competitive Profit Maximization

As before, consider a firm producing a single output y using N inputs x D

.x1 ;


; xN / according to a production function y D f .x/, and assume that the
firm is a price taker in all N factor markets. In addition, we now assume that the
firm takes output price p as given and chooses its levels of inputs to solve the fol-
lowing static competitive profit maximization problem:
N
( )
X
max pf .x/ wi xi D pf .x  / wx  : (2.1)
x0
i D1

The maximum profit   pf .x  / wx  to problem (2.1) depends on prices .w; p/

and the firm’s production function y D f .x/. The corresponding relation  D
.w; p/ between maximum profit and prices is denoted as the firm’s dual profit
function.

2.1 Properties of .w; p/

Property 2.1.
a) .w; p/ is decreasing in w and increasing in p.
b) .w; p/ is linear homogeneous in .w; p/. i.e., .w; p/ D .w; p/ for
all scalar  > 0.
c) .w; p/ is convex in .w; p/. i.e.,

 ŒwA C .1 /wB ; pA C .1 /pB   .wA ; pA /C.1 /.wB ; pB /

for all 0    1: See Figure 6.1.

11
12 2 Static Competitive Profit Maximization

Fig. 2.1 .w; p/ is convex π(w , p)

in .w; p/

d) if .w; p/ is differentiable in .w; p/, then

@.w; p/
y.w; p/ D ;
@p
(Hotelling’s Lemma)
@.w; p/
xi .w; p/ D i D 1;


; N:
@wi

Proof. Properties 2.1.a–b follow obviously from the definition of the firm’s maxi-
mization problem (2.1). In order to prove 2.1.c simply note that, for .wC ; pC / 
.wA ; pA / C .1 /.wB ; pB / and xC solving 2.1 for .wC ; pC /,

.wC ; pC /  pC f .xC / wC xC

D  pA f .xC / wA xC C .1 / pB f .xC / wB xC
   
(2.2)
 .wA ; pA / C .1 /.wB ; pB /

since .wA ; pA /  pA f .xC / wA xC , .wB ; pB /  pB f .xC / wB xC . In order to

prove Hotelling’s Lemma 2.1.d simply total differentiate .w; p/  pf .x  / wx 
with respect to .w; p/ respectively and then apply the standard first order conditions
for an interior competitive profit maximum:
N 
@f .x  / @xk

@.w; p/ 
X
D f .x / C p wk
@p @xk @p
kD1
D f .x  /
(2.3)
N 
@f .x  / @xk

@.w; p/ 
X
D xi C p wk
@wi @xk @wi
kD1
D xi i D 1;


; N:

t
u
2.2 Corresponding properties of y.w; p/ and x.w; p/ 13

Hotelling’s Lemma plays the same role in the theory of competitive profit max-
imization as Shephard’s Lemma plays in the theory of competitive cost minimiza-
tion. Hotelling’s Lemma is an envelope theorem. The Lemma applies only for in-
finitesimal changes in a price and yet is critical to the empirical theoretical applica-
tion of dual profit functions.

2.2 Corresponding properties of y.w; p/ and x.w; p/

Property 2.2.
a) y.w; p/ and x.w; p/ are homogeneous of degree 0 in .w; p/. i.e.,
y.w;
" p/ D y.w; p/ and x.w;
# p/ D x.w; p/ for all scalar   0.
@x.w;p/ @y.w;p/
@w N N @w N 1
b) @x.w;p/ @y.w;p/ is symmetric positive
@p 1N @p 11 .N C1/.N C1/
semidefinite.

1
Proof. Property 2.2.a followsh directly ifrom the maximization problem (2.1). In or-
2
der to prove 2.2.b, note that @ @w@p
.w;p/
is symmetric positive semidef-
.N C1/.N C1/
inite by 2.1.c and then apply 2.1.d to evaluate this matrix. u
t

Moreover, 2.2.a–b exhaust the (local) properties that are placed on output supply
y.w; p/ and factor demand x.w; p/ relations by the hypothesis of competitive profit
maximization (2.1).

Proof. First total differentiate .w; p/  pf .x.w; p// wx.w; p/ to obtain

N
@.w; p/ @y.w; p/ X @xk .w; p/
 y.w; p/ C p wk
@p @p @p
kD1
N
@.w; p/ @y.w; p/ X @xk .w; p/
 xi .w; p/ C p wk i D 1;


; N
@wi @wi @wi
kD1
(2.4)

Property 2.2.a implies (by Euler’s theorem)

1
Alternatively, .w; p/ homogeneous of degree one in .w; p/ implies (by Euler’s theorem)
@.w;p/
@wi
D xi .w; p/ homogeneous of degree 0.
14 2 Static Competitive Profit Maximization
N
@y.w; p/ X @y.w; p/
p C wk Dc
@p @wk
kD1
N
@xi .w; p/ X @xi .w; p/
p C wk D0 i D 1;


; N
@p @wk
kD1

and property 2.2.b states the reciprocity relations

@y.w; p/ @xk .w; p/

D
@wk @p
@xi .w; p/ @xk .w; p/
D i; k D 1;


; N;
@wk @wi
so properties 2.2.a–b jointly imply
N
@y.w; p/ X @xk .w; p/
p wk D0
@p @p
kD1
N
@y.w; p/ X @xk .w; p/
p wk D0 i D 1;


; N:
@wi @wi
kD1

Substituting this into the identity (2.4) yields

@.w; p/
D y.w; p/  0
@p
(Hotelling’s Lemma)
@.w; p/
D xi .w; p/  0 i D 1;


; N:
@wi
The reciprocity relations 2.2.b imply that the system of differential equations
@.w;p/
@p
D y.w; p/, @.w;p/
@wi
D xi .w; p/, .i D 1;


; N / integrates up to an
underlying function .w; p/ (Frobenius theorem). The positive semidefiniteness re-
striction 2.2.b implies positive semidefiniteness of the Hessian matrix of .w; p/,
and this in turn implies y D f .x/ is concave at all x  (see (2.6) below). This estab-
lishes the second order conditions on the production f .x/ for competitive profit
maximization. The first order conditions follow from the fact that 2.2 establish
Hotelling’s Lemma for all @y.w;p/
@p
, @x.w;p/
@p
, @y.w;p/
@w
, @x.w;p/
@w
satisfying properties
2.2 (see (2.3)). ut

2.3 Second order relations between .w; p/ and f .x/, c.w; y/

As in the case of a cost function, the firm’s production function y D f .x/ can
be recovered from knowledge of the profit function .w; p/. Given knowledge of
2.3 Second order relations between .w; p/ and f .x/, c.w; y/ 15

.w; p/ and differentiability of .w; p/, application of Hotelling’s Lemma imme-

diately gives a profit maximizing combination .x; y/ for prices .w; p/.
Likewise the first and second derivatives of f .x/ at x D x.w; p/ can be calcu-
lated directly from the first and second derivatives of .w; p/. The first derivatives
can be calculated simply as

@f Œx.w; p/ wi
D i D 1;


; N (2.5)
@xi p
using the first order conditions for an interior solution to problem (2.1). The second
derivatives can be calculated from the matrix equation
1
@2 f .x.w; p// @2 .w; p/
   
p˝ D (2.6)
@x@x N N @w@w N N
h 2 i
assuming an inverse for @ @w@w
.w;p/
.
/
Proof. Simply total differentiate the first order conditions p @f@x
.x
i
wi D 0 (i D
1;


; N ) with respect to w to obtain
N
X @2 f .x  / @xk .w; p/
p 1D0 i; j D 1;


; N; (2.7)
@xi @xk @wj
kD1

2
Substitute @xk@w
.w;p/
j
D @@w.w;p/
k @wj
(by Hotelling’s Lemma) into (2.7) and express
the result in matrix form. t
u
This result (2.6) can easily be extended to the case of multiple outputs (Lau
1976).
Since elasticities of substitution (holding output y constant) and scale effects
are easily expressed in terms of a dual cost function c.w; y/, it is useful to note that
.w; p/ also provides a second order approximation to c.w; y/. The first derivatives
of c.w; y/ can be calculated simply as

@c.w; y  /
D xi .w; y  /
@wi
@.w; p/
D i D 1;


; N (2.8)
@wi
@c.w; y  /
Dp
@y

where y   y.w; p/ D @.w;p/ @p

, the profit maximizing level of output given
.w; p/. The second derivatives of c.w; y/ at y  D y.w; p/ can behcalculatedi from
2 c.w;y/
.w; p/ using the following matrix relations, here cww .w; y/  @ @w@w ,
h 2 i h 2 i N N
@ c.w;y/ @ .w;p/
cwy .w; y/  @w@y
, ww .w; p/  @w@w
, wp .w; p/ 
N N N N
16 2 Static Competitive Profit Maximization
h i
@2 .w;p/
@w@p
,
N N

cww .w; y  / D ww .w; p/ C wp .w; p/ pp .w; p/ 1

wp .w; p/T
cwy .w; y  / D wp .w; p/ pp .w; p/ 1
(2.9)
 1
cyy .w; y / D pp .w; p/

Proof. xi .w; p/ D xi Œw; y.w; p/ .i D 1;


; N / implies (by Hotelling’s Lemma

and Shephard’s Lemma)

wi .w; p/ D cwi .w; y  / (2.10)

where y  D y.w; p/. Differentiating with respect to .w; p/,

ww .w; p/ D cww .w; y  / C cwy .w; y  / wp .w; p/T

(2.11)
wp .w; p/ D cwy .w; y  / pp .w; p/

Combining (2.11),

cww .w; y  / D ww .w; p/ C wp .w; p/ pp .w; p/ 1

wp .w; p/T
(2.12)
cwy .w; y  / D wp .w; p/ pp .w; p/ 1

Finally differentiating the first order condition cy .w; y  / D p (for profit maximiza-
tion) with respect to p yields cyy .w; y  / D pp .w; p/ 1 .

2.4 Additional properties of .w; p/

Property 2.3.
a) the partial elasticity of substitution ij between inputs i and j , allowing for
variation in output, can be defined as
2
.w; p/ @@w.w;p/
i @wj
ij .w; p/ D @.w;p/ @.w;p/
i; j D 1;


; N
@wi @wj

and, in the case of multiple outputs y D .y1 ;


; yM /, the partial elasticity

of transformation between outputs i and j can be defined as
2
.w; p/ @ @p.w;p/
i @pj
tij .w; p/ D @.w;p/ @.w;p/
i; j D 1;


; N
@pi @pj
2.5 Le Chatelier principles and restricted profit functions 17

b) the transformation function f .x; y/ D 0 is disjoint in outputs only if

@2 .w; p/
D0 for all i ¤ j , all .w; p/:
@pi @pj

2.5 Le Chatelier principles and restricted profit functions

Samuelson proved the following Le Chatelier principle: fixing an input at its initial
static equilibrium level dampens own-price comparative static responses, or more
precisely

@y.w; p/ @y.w; p; xN 1 /
 0
@p @p
(2.13)
@xi .w; p/ @xi .w; p; xN 1 /
 0
@wi @wi

where xN 1  x1 .w; p/, i.e., @y.w;p/

@p
, @x.w;p/
@w
denote the comparative static changes
in .y; x/ when the level of input 1 cannot vary from its initial equilibrium level
x1 .w; p/ (Samuelson 1947). The following generalization of this result is easily
established using duality theory:
" # " #
@x.w;p/ @y.w;p/ @x.w;p;xN 1 / @y.w;p;xN 1 /
@w @w @w @w
@x.w;p/ @y.w;p/ @x.w;p;xN 1 / @y.w;p;xN 1 /
@p @p .N C1/.N C1/ @p @p .N C1/.N C1/
(2.14)
is a positive semidefinite matrix.

Proof. By the definition of competitive profit maximization (2.1),

N N
( ) ( )
X X
.w; p/  max pf .x/ wi xi  max pf .x/ wi xi  .w; p; x1A /
x0 x0
i D1 i D1

s.t. x1 D x1A
(2.15)
i.e., adding a constraint x1 D x1A to a maximization problem (2.1) generally de-
creases (and never increases) the maximum attainable profits. Then

 0 for all .w; p; x1A /

(
A A
.w; p; x1 /  .w; p/ .w; p; x1 / (2.16)
D 0 for x1A D x1 .w; p/

i.e., .w; p; x1A / attains a minimum (D 0) over .w; p/ at all x1A D x1 .w; p/. By the
second order condition for an interior minimum,
18 2 Static Competitive Profit Maximization

@2 .w; p; x1A / @2 .w; p/ @2 .w; p; x1A /

     
 (2.17)
@w@p .N C1/.N C1/ @w@p @w@p

is positive semidefinite at x1A D x1 .w; p/. (2.17) and Hotelling’s Lemma establish
(2.14). u t

Samuelson’s Le Chatelier Principle (2.13)/(2.14) has often been given the fol-
lowing dynamic interpretation: assuming that the difference between short-run, in-
termediate run and long-run equilibrium can be characterized in terms of the number
of inputs that can be adjusted within these time frames, the magnitude of the firm’s
response xi (or y) to a given change in price wi (or p) increases over time, and
the sign of these responses does not vary with the time frame.
However this characterization of dynamics in terms of a series of static models
with a varying number of fixed inputs is unsatisfactory. In general dynamic behavior
must be analyzed in terms of truly dynamic models.
For example, it often appears that an increase in price for beef output leads to a
short-run decrease in beef output and a long-run increase in output, which contra-
dicts the dynamic interpretation of Samuelson’s Le Chatelier Principle (2.13). This
can be explained in terms of the dual role of cattle as output and as capital input
to future production of output: a long-run increase in output generally requires an
increase in capital stock, and this can be achieved by a short-run decrease in output
(Jarvis 1974). This illustrates the following point: a series of static models with a
varying number of fixed inputs completely ignores the intertemporal decisions as-
sociated with the accumulation of capital (durable goods).
Nevertheless, versions of profit functions conditional on the levels of certain in-
puts can be useful in applied work. The following restricted dual profit function is
conditional on the level of capital stocks K:
N
( )
X
.w; p; K/  max pf .x; K/ wi xi : (2.18)
x0
iD1

.w; p; K/ has the same properties in its price space .w; p/ as does the unrestricted
profit function .w; p/, which implcitly treats capital (or services from capital) as
a freely adjustable input. In addition @.w;p;K/
@K
measures the shadow price of cap-
ital, and twice differentiability of .w; p; K/ and Hotelling’s Lemma establish the
following reciprocity conditions:
 
@y.w; p; K/ @ @.w; p; K/
D
@K @p @K
  (2.19)
@xi .w; p; K/ @ @.w; p; K/
D i D 1;


; N:
@K @wi @K

There are three major advantages to specifying a restricted dual profit function.
First, a restricted profit function .w; p; K/ is consistent with short-run equilibrium
for a variety of dynamic models that dichotomize inputs as being either perfectly
2.6 Application of dual profit functions in econometrics: I 19

variable or quasi-fixed in the short-run. Second, it is more realistic to assume that

firms are in a short-run equilibrium rather than in a long-run equilibrium, and mis-
specifying an econometric model as long-run equilibrium (e.g. using an unrestricted
profit function .w; p/) implies that the resulting estimators of long-run equilibrium
effects of policy are unreliable. Third, long-run equilibrium effects can sometimes
be inferred correctly from the estimates of .w; p; K/ using the following first order
condition for a long-run equilibrium level of capital stock K  :

@.w; p; K  / @.w; p; K  /
D wK or D .r C ı/pK (2.20)
@K @K
where wK  rental price of capital, pK  asset price of capital, r  an appropriate
discount rate and ı  rate of depreciation for capital. After estimating .w; p; K/
we would solve (2.20) for K  .2

2.6 Application of dual profit functions in econometrics: I

As in the use of a cost function c.w; y/, the above theory is usually applied by first
specifying a functional form .w; p/ for a profit function .w; p/ and differentiat-
ing .w; p/ with respect to .w; p/ to obtain the estimating equations

@ .w; p/
yD
@p
(2.21)
@ .w; p/
xi D i D 1;


; N
@wi
@y
(employing Hotelling’s Lemma). Then the symmetry restrictions @w i
D @x @p
i
,
h 2 i
@xi @xj @
@wj
D @wi , (i; j D 1;


; N ) are tested and @w@p is checked for
.N C1/.N C1/
positive semidefiniteness. For example, a profit function could be postulated as
N X
N N
X p p X p p
D aij wi wj C a0j wj p C ap
i D1 j D1 j D1

2 @.w;p;K/
In general @K
cannot be measured directly from the estimating equations (a) y D
@.w;p;K/ @.w;p;K/ P P p p
@p
, (b) x D @w
, e.g. when  D K i j aij vi vj C aKK K,
(v  .w; p/). Estimates of @.w;p;K/
@K
can be obtained either by estimating  D .w; p; K/
jointly with (a)–(b) and then calculating directly @.w;p;K/
@K
, or by estimating (a)–(b) and then
PN @xi .w;p/
calculating @.w;p/
@K
D p @y.w;p/
@K
w
i D1 i @K
. This last equation is derived as fol-
lows, .w; p; K/ D .w; p; K/ ) @.w;p;K/ @K
D  @.w;p;K/
@K
) @.w;p;K/
@K
D
2
@ .w;p;K/ P N 2
@ .w;p;K/
p @p@K C i D1 wi @wi @K (Euler’s theorem), and then applying (2.19).
20 2 Static Competitive Profit Maximization

(this is a Generalized Leontief functional form, which imposes homogeneity of de-

gree one in prices on the profit function). This leads to the following equations for
estimation
N   12
X wj
y DaC a0j
p
j D1
 12  12 (2.22)
N  
X wj p
xi D aij C ai 0 i D 1;


; N
wi wi
j D1

(applying Hotelling’s Lemma). Here the symmetry restrictions are expressed as

aij D aj i (i; j D 1;


; N ) and a0i D ai 0 (i D 1;


; N ). If and only if
the symmetry and second order conditions are satisfied, then equations (2.21) can
be interpreted as being derived from a profit function .w; p/ for a produces show-
ing static, competitive profit maximizing behavior. Similar comments apply to the
estimation of a restricted profit function .w; p; K/.
Here we can note three major advantages of this approach to the modeling of
producer behavior. First, it enables us to specify systems of output supply and factor
demand equations that are consistent with profit maximization and with a general
specification of technology. This cannot be achieved by estimating a production
function directly together with first order conditions for profit maximization. Sec-
ond, modeling a dual profit function explicitly allows for the endogeneity of output
levels to the producer. This is in contrast to cost functions c.w; y/ where output
generally is treated as exogenous. Third, aggregation problems are likely to be less
severe for unrestricted dual profit functions than for dual cost functions. If all firms
f D 1;


; F (a constant number of firms over time) face identical prices .w; p/
then it is clear that a profit function can be well defined for aggregate market data:
F
X
f .w; p/ D ˆ.w; p/  ….w; p/: (2.23)
f D1

In the special case of profit maximization at identical prices, a cost function also is
well defined for data aggregated over firms even though the output level yf varies
over firms (we shall use this in a later lecture); but in this case we may as well
estimate ….w; p/ directly.
The disadvantage of estimating a profit function .w; p/ rather than a cost func-
tion c.w; y/ is that the former imposes stronger behavioral assumptions which are
often very unrealistic. For example, at the time of production decisions, farmers gen-
erally have better knowledge of input prices than of output prices forthcoming at the
time of marketing in the future. Thus risk aversion and errors in forecasting prices
are more likely to influence the choice of output levels rather than to contradict the
hypothesis of cost minimization. In addition in the case of food retail industries,
hypothesis of oligopoly behavior plus competitive cost minimization may be more
realistic than the hypothesis of competitive profit maximization.
2.7 Industry profit functions and entry and exit of firms 21

2.7 Industry profit functions and entry and exit of firms

As mentioned above, an industry profit function is well defined provided that all
firms face identical prices .w; p/ and the composition of the industry in terms of
firms does not change. In this case the industry profit function is simply the sum of
the profit functions of the firms f D 1;


; F : ….w; p/ D fFD1 f .w; p/ for all
P
.w; p/.
A more realistic assumption is that changes in prices .w; p/ induce some estab-
lished firms to exit the industry and some new firms to enter the industry. Suppose
that there is free entry and exit to the industry (all firms in the industry earn non-
negative profits because all firms that would earn negative profits at .w; p/ are able
to exit the industry, and all firms excluded from the industry are also at long-run
equilibrium). Then the industry profit function inherits essentially the some proper-
ties as the individual firm’s profit function f .w; p/.
Proposition 2.1 provides a characterization of the industry profit function .w; p/,
assuming (a) competitive behavior, (b) both input prices w D .w1 ;


; wN / and
output prices p are exogenous, and (c) a continuum of firms and free entry/exit to
the industry.
The role of the assumption of a continuum of firms deserves comment. As noted
by Novshek and Sonnenschein (1979) in the case of marginal consumers, an in-
finitesimal change in price will lead to entry/exit behavior only in the case of a
continuum of agents. Therefore it is necessary to assume that there exists a contin-
uum of firms. Industry profits are calculated by integrating over the continuum of
firms in the industry:
Z f m .w;p/
….w; p/ D .w; p; f /.f / df
1

where .f / is the density of firms f . .w; p; f / denotes the individual firm’s profit
function conditional on the firm being in the industry, and industry profits are ob-
tained by integrating over those firms in the industry given prices .w; p/ and free
entry/exit.
Adapting the arguments of Novshek and Sonnenschein, the assumption of a con-
tinuum of firms also establishes the differentiability of the industry profit function
….w; p/, industry factor demands X.w; p/ and industry output supplies Y .w; p/
with free entry/exit. The argument can be outlined as follows. Since the individual
firm’s profit function .w; p; f / is conditional on firm f remaining in the indus-
try, it is reasonable to assume that .w; p; f / is twice differentiable in .w; p/, i.e.,
.w; p; f / is not kinked at  D 0 due to exit from the industry. .w; p; f / is dif-
ferentiable in f and the derivative f .w; p; f / < 0 assuming a continuum of firms
indexed in descending order of profits. Then (by the implicit function theorem) a
marginal firm f m D f m .w; p/ is defined implicitly by the zero profit condition
.w; p; f m / D 0, and f m .w; p/ is differentiable. Under these assumptions it can
easily be shown that the industry profit function with free entry/exit is differentiable,
and its derivatives …w .w; p/, …p .w; p/ can be calculated by applying Leibnitz’s
22 2 Static Competitive Profit Maximization

rule to the above equation for industry profits (note that the zero profit condition
.w; p; f m / D 0 established Hotelling’s Lemma at the industry level). A similar
procedure establishes the differentiability of industry factor demands X.w; p/ and
output supplies Y .w; p/ with free entry/exit.

Proposition 2.1. Assume that the industry consists of a continuum of firms such that
each individual firm’s profit function .w; p; f / is twice differentiable in .w; p/,
differentiable in f , f .w; p; f / < 0, and is linear homogeneous and convex
in .w; p/ and satisfies Hotelling’s Lemma. Also assume .w; p; f m / D 0 for a
marginal firm f m . Then .w; p/ is linear homogeneous and convex in .w; p/.
Moreover, it also satisfies Hotelling’s Lemma, i.e.,

@….w; p/
Xi .w; p/ D i D 1;


; N
@wi
(2.24)
@….w; p/
Yk .w; p/ D k D 1;


; N
@pk

and industry derived demands X.w; p/ and output supplies Y .w; p/ are differen-
tiable.

Proof. Given prices .w; p/ and free entry/exit, index firms in the industry in de-
scending order of profits. Industry profits can be calculated as
Z f m .w;p/
….w; p/ D .w; p; f / .f / df (2.25)
1

where .f / is the density of firms f and a marginal firm f m satisfies the zero profit
condition
.w; p; f m / D 0: (2.26)
Assuming .w; p; f / differentiable in .w; p; f / and f .w; p; f / < 0, the zero
profit condition (2.26) establishes (using the implicit function theorem) f m D
f m .w; p/ is defined and differentiable. Differentiability of .w; p; f / and f m .w; p/
establishes (using (2.25) ….w; p/ is differentiable). Applying Leibnitz’s rule to
(2.25), Pro:2.1

f m .w;p/
@….w; p/ @.w; p; f / @f m .w; p/
Z
D .f / df C  .w; p; f m .w; p//  .f m .w; p//
@wi 1 @wi @wi
D Xi .w; p/ i D 1;


; N
(2.27)
Z f m .w;p/
@….w; p/ @.w; p; f / @f m .w; p/
D .f / df C  .w; p; f m .w; p//  .f m .w; p//
@pk 1 @pk @pk
D Yk .w; p/ k D 1;


; M
(2.28)
2.8 Application of dual profit functions in econometrics: II 23

using Hotelling’s Lemma for the individual firm in the industry and the marginal
condition  .w; p; f m .w; p// D 0. Similarly
Z f m .w;p/
xi .w; p/ D xi .w; p; f / .f / df i D 1;


; N (2.29)
1
Z f m .w;p/
yk .w; p/ D yk .w; p; f / .f / df k D 1;


; M (2.30)
1

Differentiability of x.w; p; f /, y.w; p; f /, f m .w; p/ implies (using (2.29)–(2.30))

X.w; p/, Y .w; p/ differentiable. ….w; p/ homogeneous of degree one in .w; p/
follows from (2.25): .w; p; f / is linear homogeneous in .w; p/ for all f , and
the continuum of firms 1;


; f m .w; p/ is invariant to equiproportional changes
in .w; p/. Convexity of ….w; p/ can be established as follows. Define a profit
N
function .w; p; g/ for each firm g allowing for free entry/exit to the industry:
N
.w; p; g/  0 for all .w; p; g/, and .w;
N p; g/ D 0 when the firm has exited the
industry in order to avoid negative profits. Thus ….w; p/ with entry/exit is the sum
N
or integral over the continuum of profit functions .w; p; g/ for the fixed set of po-
N
tential firms. .w; p; g/ is convex in .w; p/ by standard arguments (e.g. McFadden
1978), and in turn ….w; p/ is convex in .w; p/. u t

2.8 Application of dual profit functions in econometrics: II

Our development of Hotelling’s Lemma when the number of firms is variable to

the industry has important implications for empirical studies. An industry profit
function ….w; p/ satisfies Hotelling’s Lemma in two extreme cases: the number
of firms in the industry is fixed (the standard case) or there is free entry/exit to
the industry (Proposition 2.1). In intermediate cases, where the number of firms is
variable but entry/exit is not instantaneous and costless, the Lemma does not apply.
This can be seen from equations (2.29) and (2.30) in the proof of Proposition 2.1:
applying Leibnitz’s rule to the integral for industry profits over a continuum of firms
with entry/exit
Z f m .w;p/
….w; p/ D .w; p; f /.f / df
1
yields Hotelling’s Lemma only if there exist marginal firms earning zero profits, i.e.,
.w; p; f m .w; p// D 0 (assume for the sake of argument that f m .w; p/ is differ-
entiable in the intermediate case). This zero profit condition is characteristic of free
entry/exit with a continuum of firms but it is not characteristic of the intermediate
case.
The usual assumptions that are acknowledged in standard applications of Hotelling’s
Lemma to industry-level data are that each firm in the industry shows static com-
petitive profit maximizing behavior (conditional on the firm being in the industry).
In addition we must generally add the restrictive assumption that the composition
24 2 Static Competitive Profit Maximization

of the industry does not change over the time period of the data or there is free
entry/exit to the industry.
Nevertheless there is at least in principle a simple procedure for avoiding this ad-
ditional restrictive assumption: the industry profit function can be defined explicitly
as conditional upon the number of firms of each different type. For example, suppose
that an industry consists of two homogeneous types of firms in variable quantities
F1 and F2 . The industry profit function can be written as ….w; p; F1 ; F2 /, where
F1 nd F2 are specified as parameters along with .w; p/. Then X.w; p; F1 ; F2 / D
…w .w; p; F1 ; F2 /, Y .w; p; F1 ; F2 / D …p .w; p; F1 ; F2 / by standard argument
(e.g. Bliss). Of course in practice reasonable data on the number of firms by type
is not always available, but whenever possible such modifications seem likely to
improve the specification of the model.
Thus, if we have time series data on the number of firms F1 and F2 in the two
classes as well as data on total output, total inputs and prices, then we can postu-
late an industry profit function ….w; p; F1 ; F2 / conditional on F1 , F2 and apply
Hotelling’s Lemma to obtain the estimating equations

@….w; p; F1 ; F2 /
Y D
@p
(2.31)
@….w; p; F1 ; F2 /
Xi D i D 1;


; N:
@wi
If there is free entry/exit to the industry, then (by 2.1) parameters F1 and F2 drop
out of the system of estimating equations (2.31)—in this manner the assumption
of long-run industry equilibrium is easily tested. If there is not free entry/exit and
if the number of firms F1 and F2 varies over time, then the parameters F1 and F2
are significant in equations (2.31). The stocks of firms F1 and F2 at any time t
probably can be approximated as predetermined at time t (i.e., the number of firms
is essentially inherited from the past, given substantial delays in entry and exit).
Then F1;t and F2;t do not necessarily covary with the disturbance terms at time t
for equations (2.31), so that equations (2.31) may be estimated consistently even if
there is costly entry/exit to the industry.
However for policy purposes it may be desirable to estimate (2.31) jointly with
equations of motion for the number of firms:

F1;tC1 F1;t D F1 .w t ; p t ; F1;t ;


/

(2.32)
F2;tC1 F2;t D F2 .w t ; p t ; F2;t ;


/

Equations (2.31) can indicate the short-run impact of price policies or industry out-
put and inputs levels .Y; X /, i.e., the impact of the price policies before there is an
adjustment in the number of firms.
Equations (2.31) and (2.32) jointly indicate intermediate and long-run impacts of
price policies. For example, at a static long-run equilibrium there is no exit/entry to
the industry, so the long-run equilibrium numbers of firms F1 , F2 for any .w; p/

can be calculated from (2.32) by solving the implicit equations F1 .w t ; p t ; F1;t /D
References 25


0, F2 .w t ; p t ; F2;t / D 0. Obvious difficulties here are (a) problems in specifying
the dynamics of entry and exit (equations (2.32)) correctly, and (b) dangers in using
(2.31) to extrapolate to long-run equilibrium numbers of firms F1 , F2 that are
outside of the data set.

References

1. Fuss, M.& McFadden, D. (1978). Production Economics: A Dual Approach to Theory and
Applications: The Theory of Production, History of Economic Thought Books, McMaster
University Archive for the History of Economic Thought
2. Jarvis, L. (1974). Cattle as Capital Goods and Ranchers as Portfolio Managers: An Applica-
tion to the Argentine Cattle Sector, Journal of Political Economy, pp. 489–520
3. Lau, L. (1976). A Characterization of the Normalized Restricted Profit Function, Journal of
Economic Theory, pp. 131–163.
4. Novshek, W. & Sonnenschein, H .(1979). Supply and marginal firms in general equilibrium,
Economics Letters, Elsevier, vol. 3(2), pp. 109–113.
5. Samuelson, P. A. (1947). Enlarged ed., 1983. Foundations of Economic Analysis, Harvard
University Press.
Chapter 3
Static Utility Maximization and Expenditure
Constraints

Here we model both consumer and producer behavior subject to expenditure con-
straints. We begin with the case of consumer.
Consider a consumer maximizing utility u by allocating his income y among
N commodities x D .x1 ;


; xN /, and denote his utility function as u D u.x/.
Assume that the consumer takes commodity prices p D .p1 ;


; pN / as given and
solves the following static competitive utility maximization problem:

max u.x/ D u.x  /

x0
N
X (3.1)
s.t. pi xi  y
i D1

The maximum utility V  u.x  / to problem (3.1) depends on prices and income
.p; y/ and the consumer’s utility function u D u.x/. The corresponding relation
V D V .p; y/ between maximum utility and prices and incomes is denoted as the
consumer’s dual indirect utility function.
A necessary condition for utility maximization (3.1) is that the consumer attains
the utility level u.x  / at a minimum cost. In other words, if x  does not minimize
the cost N D u.x  /, then a higher utility level u > u.x  /
P
i D1 pi xi subject to u.x/
PN
can be attained at the same cost i D1 pi xi D y (for this result we only need to
assume local nonsatiation of u.x/ in neighborhood of x  ). Thus a solution x  to
the utility maximization problem (3.1) also solves the following cost minimization
problem when the exogenous utility level u is equal to u.x  /:
N
X N
X
min pi xi D pi xi
x0
i D1 i D1
(3.2)
s.t. u.x/  u

The minimum cost E D N 

P
i D1 pi xi to problem (3.2) depends on prices and utility
level .p; u/ and the consumers utility function u.x/. The corresponding relation

27
28 3 Static Utility Maximization and Expenditure Constraints

between minimum expenditure and prices and utility level, E D E.p; u/, is denoted
as the consumers dual expenditure function. Note that (3.2) is formally equivalent to
the producer’s cost minimization problem minx0 N
P
iD1 wi xi s.t. f .x/  y (1.1).
Thus the expenditure function E.p; u/ inherits the essential properties (1.1) of the
producers cost function c.w; y/:

Property 3.1.
a) E.p; u/ is increasing .p; u/.
b) E.p; u/ is linear homogeneous in p.
c) E.p; u/ is concave in p.
d) If E.p; u/ is differentiable in p, then

@E.p; u/
xi .p; u/ D i D 1;


; N: (Shephard’s Lemma)
@pi

Also note that a sufficient condition for utility maximization (3.1) is that the
consumer attains the utility level u.x  / at a minimum cost equals to N 
P
i D1 pi xi .
A A A
In other words, if x solves (3.2) subject to u.x/ D u , then x also solves (3.1)
subject to N
P PN A
i D1 pi xi D y  iD1 pi xi .

Proof. Assume the that u.x/ is continuous and x A solves (3.2) at the exogenously
determined utility level u  uA . Now suppose that x  rather than x A solves (3.1) at
the exogenously determined expenditure level y A  N A
P
i D1 pi xi > 0. This would
imply u.x  / > u.x A / and (given local nonsatiation) iD1 pi xi D N
N A
P P
iD1 pi xi .
Then continuity of u.x/ would imply that there exits an xQ in the neighborhood of x 
Q  u.x A / and N
such that u.x  /  u.x/ Qi < N A
P P
i D1 pi x i D1 pi xi , which contradicts
the assumption x solves (3.2). Therefore x solves (3.2) implies x A solves (3.1).
A A

t
u
Thus x A solves (3.1) if and only if x A solves (3.2) subject to u  u.x A /. This
implies that the restrictions placed on Marshallian consumer demands x D x.p; y/
(corresponding to problem (3.1)) by the hypothesis of utility maximization (3.1) can
be analyzed equivalently in terms of the restrictions placed on Hicksian consumer
demands x D x h .p; u/ (corresponding to problem (3.2)) by the hypothesis of cost
minimization (3.2). In other words, the hypothesis of cost minimization (3.2) ex-
hausts the restrictions placed on Marshallian demands x D x.p; y/ by the hypoth-
esis of utility maximization. Properties 3.1 of E.p; u/ imply

Property 3.2.
a) x.p; u/ are homogenous of degree 0 in p. i.e. x.p; u/ D x.p; u/ for all
scalar  > 0.
3.1 Properties of V .p; y/ 29

 
@x.p; u/
b) is symmetric negative semidefinite.
@p N N

And properties 3.2 exhausts the implications of cost minimization for the (local)
properties of Hicksian demands x h .p; u/ (the proof is the same as in the case of
cost minimization by a producer). Therefore, by the proof immediately above, 3.2
also exhausts the implications of utility maximization (3.1) for the (local) properties
of Marshallian demands x.p; y/, where 3.2 is evaluated at a utility maximization
u D u .x.p; y//. Of course the characterization of utility maximization in terms
of (3.2) is not immediately useful empirically in the sense that utility level u is not
observed (nor is u exogenous to the consumer).
This relation between utility maximization and cost minimization is very differ-
ent from the relation between profit maximization and cost minimization for the
producer: profit maximization implies but is not equivalent to cost minimization
in any sense. The explanation is that in the consumer case, in contrast to the pro-
ducer case, maximization is subject to an expenditure constraint defined over all
commodities.

3.1 Properties of V.p; y/

Property 3.3.
a) V .p; y/ is decreasing in p and increasing in y.
b) V .p; y/ is homogenous of degree 0 in .p; y/. i.e. V .p; y/ D V .p; y/ for
all scalar  > 0.
c) V .p; y/ is quasi-convex in p. i.e. fp W V .p; y/  kg is a convex set for all
scalar k  0. See Figure 3.1
d) If V .p; y/ is differentiable in .p; y/, then

@V .p; y/=@pi
xi .p; y/ D i D 1;


; N
@V .p; y/=@y
(Roy’s Theorem)

Proof. Properties 3.3.a–b follows simply from the definition of the consumer’s max-
imization problem (3.1). For a proof of 3.2.c see Varian (1992, pp. 121–122). In or-
der to prove 3.3.d note that, if u is the maximum utility for (3.1) given parameters
.p; y/, then u  V .p; y  / where y   E.p; u /, i.e. y  is the minimum expen-
diture necessary to attain a utility level u given prices p. Total differentiating this
30 3 Static Utility Maximization and Expenditure Constraints

Fig. 3.1 V .p; y/ is quasi-

convex in p

identity u  V .p; E.p; u // with respect to pi ,

@V .p; y  / @V .p; y  / @E.p; u /

0D C i D 1;


; N (3.3)
@pi @y @pi

which yields 3.3.b since @E.p; u /=@pi D xih .p; u / D xi .p; y  /, i D 1;


; N .

t
u

Roy’s Theorem and V .p; y/ quasi-convex in p can also be derived as follows.

Since V .p; y/  maxx u.x/ s.t. px D y,

V .p; y/ u.x/  0 for all .p; y; x/ such that px D y (3.4)

or equivalently
 P 
V p; N
i D1 pi xi u.x/  0 for all .p; x/ (3.5)

where y is now defined implicitly as y D N

P
i D1 pi xi . By (3.5),
 P 
G.p; x/  V p; N i D1 pi xi u.x/  0 for all p
 P 
G.p; x/  V p; N i D1 pi xi u.x/ D 0 at p such that x D x.p; y/ (3.6)

for y  N
P
iD1 pi xi

In other words, given that x A solves (3.1) conditional on .p A ; y A /, then G.p; x A /

attains a global minimum over p at p A . This implies the following first and second
order conditions for maximization:
@G.p A ; x A /
D0 i D 1;


; N
@pi
 2 (3.7)
@ G.p A ; x A /

is symmetric positive semidefinit.
@p@p N N
 P 
Since G.p; x/  V p; N iD1 pi xi u.x/, conditions (3.7) imply
3.2 Corresponding properties of x.p; y/ solving problem (3.1) 31

@G.p; x/ @V .p; y/ @V .p; y/

 C xi D 0 i; j D 1;


; N (3.8a)
@pi @pi @y
@2 G.p; x/ @2 V .p; y/ @2 V .p; y/ @2 V .p; y/ @2 V .p; y/
 C xj C xi C xj xi
@pi @pj @pi @pj @pi @y @y@pi @y@y
(3.8b)
h i
@2 G.p;x/
(3.8a) is Roy’s Theorem. @p@p
symmetric positive semidefinite (3.7) im-
N N
plies semidefinite—these are the second order restrictions implied by V .p; y/ quasi-
convex in p.

3.2 Corresponding properties of x.p; y/ solving problem (3.1)

Property 3.4.
a) x.p; y/ is homogeneous of degree 0 in .p; y/. i.e. x.p; y/ D x.p; y/ for
all scalar   0.
@xi .p; y/ @x h .p; u / @xi .p; y/
b) D i xj .p; y/ i; j D 1;


; N
@pj @pj @y h i
@x h .p;u /
(Slutsky equation) where u  V .p; y/ and @p
is symmetric
N N
negative semidefinite.

Proof. Property 3.4.a is obviously true. In order to prove 3.4.b, let u be the max-
imal utility for problem (3.1) conditional on .p; y/. Then we have proved the fol-
lowing identity (see pages 27–28): xi .p; y/  xih .p; u / where y  E.p; u /,
i.e.
xih .p; u /  xi p; E.p; u /


i D 1;


; N (3.9)
In words, the Hicksian and Marshallian demands x h .p; u/ and x.p; y/ are equal
when x h .p; u/ are evaluated at a utility level u D u solving (3.1) for prices p and
a given income y and x.p; y/ are evaluated at p and a level y  E.p; u / solving
(3.2) given p and the level u D u . Differentiating (3.9) with respect to pj (holding
utility level u constant) yields

@xih .p; u / @xi .p; y/ @xi .p; y/ @E.p; u /

D C i; j D 1;


; N: (3.10)
@pj @pj @y @pj
 
Since (using Shephard’s
h iLemma) @E.p; u /=@pj D xj .p; u / D xj .p; y/ .j D
@x.p;u /
1;


; N / and @p
is symmetric negative semidefinite from 3.2.b, (3.10)
N N
establishes 3.4.b. u t
32 3 Static Utility Maximization and Expenditure Constraints

The integrability problem has traditionally been prosed as follows: given a set of
Marshallian consumer demand relations x D x.p; y/, what restrictions on x.p; y/
exhaust the hypothesis of competitive utility maximizing behavior by the consumer?
We can now construct an answer as follows. Since utility maximization (3.1) and ex-
penditure minimization subject to u.x/ D u are equivalent (see pages 27–28), we
can rephrase the question as: what restrictions on x.p; y/ exhaust the hypothesis of
competitive cost minimizing behavior by the consumer? h Using thei Slutsky equation
@x h .p;u /
3.4.b we can recover the matrix of substitution effects @p
for the cost
N N
minimization demands x h .p; u / from the Marshallian
h demands
h 
i x.p; y/. x .p; u /
@x h .p;u /
homogeneous of degree 0 in p and symmetry of @p
imply the set
N N
of differential equations xih .p; u / D @E.p; u /=@pi , i D 1;


; N (Shephard’s
Lemma) (see page 4). By the Frobenius theorem, these differential
h h equations
i can be
/
integrated up to a macrofunction E.p; u / if and only if @x .p;u@p
is sym-
h h i N N

metric. Furthermore @x .p;u@p
/
negative semidefinite implies (by Shephard’s
h 2 i N N

Lemma) @ E.p;u@p@p
/
negative semidefinite, which in turn implies E.p; u /
N N
concave and u.x/ quasi-concave at x.p; y/. Thus Marshallian demand relations
x D x.p; y/ can be interpreted as being derived from competitive utility maximiz-
ing behavior if and only if
" #
@x h .p; u /
   
@x.p; y/ @x.p; y/
C Œx.p; y/1N 
@p N N @y N 1 @p
N N
is symmetric negative semidefinite,
(3.11a)
x.p; y/ homogenous of degree 0 in .p; y/.1 (3.11b)

3.3 Application of dual indirect utility functions in econometrics

The above theory is usually applied by first specifying a functional form .p; y/
for the indirect utility function V .p; y/ and differentiating .p; y/ with respect to
.p; y/ in order to obtain the estimating equations

@ .p; y/=@pi
xi D i D 1;


; N (3.12)
@ .p; y/=@y

1
x.p; y/ homogenous of degree 0 in .p; y/ implies x h .p; u / homogenous of degree 0 in p,
since x.p; y/  x h .p; V .p; y//.
3.3 Application of dual indirect utility functions in econometrics 33

(using Roy’s Theorem). .p; y/ is usually specified as being homogeneous of de-

gree 0 in .p; y/.
The symmetry conditions

@2 .p; y/ @2 .p; y/ @2 .p; y/ @2 .p; y/

D ; D i; j D 1;


; N
@pi @pj @pj @pi @pi @y @y@pi
(3.13)
are satisfied if and only if corresponding Hicksian demand relations xih .p; u / D
@E.p; u /=@pi .i D 1;


; N / integrate up to a macrofunction E.p; u /.
Proof. By the Frobenius theorem, the Hicksian demands xih .p; u / D @E.p; u /=@pi
.i D 1;


; N / integrate up to a macrofunction if and only if xih .p; u /=@pj D
xjh .p; u /=@pi .i; j D 1;


; N /. Differentiating xi D @V .p;y/=@pi
@V .p;y/=@y
(Roy’s Theo-
rem) with respect to .p; y/ and substituting into the Slutsky equation 3.4.b yields

@xih .p; u / Vp p Vy Vypj Vpi Vpj

 
Vp y Vy Vyy Vpi
D i j C i
(3.14)
@pj Vy Vy Vy Vy Vy
     
where Vpi pj  @p@j @V@p
.p;y/
i
, Vypj
 @
@pj
@V .p;y/
@y
, V p p
j i
 @
@pj
@V .p;y/
@pi
,
etc. Inspection of (3.14) shows that @xih .p; u /=@pj D @xjh .p; u /=@pi if and only
if Vpi pj D Vpj pi , Vypi D Vpi y .i; j D 1;


; N /. u
t
Thus the hypothesis of competitive utility maximization is verified by (a) testing
for the symmetry restrictions (3.13) and (b) checking the second order conditions
(3.8b) at all data points .p; y/ or (equivalently) using Slutsky relations to recover
the Hicksian matrix @x h =@p and checking for negative semidefiniteness.
For example an indirect utility function could be postulated as having the func-
tional form V D y= i j aij pi1=2 pj1=2 (a Generalized Leontief reciprocal in-
P P
direct utility function with homotheticity). Applying Roy’s Theorem leads to the
following functional form for the consumer demand equations:

y jND1 aij .pj =pi /1=2

P
x i D PN PN 1=2 1=2
i D 1;


; N: (3.15)
j D1 kD1 aj k pj pk

Here the symmetry restrictions (3.13) are expressed as aij D aj i .i; j D 1;


; N /

which are easily tested. Equations (3.15) can be interpreted as being derived from an
indirect utility function V .p; y/ D y= i j aij pi1=2 pj1=2 for a consumer show-
P P
ing static competitive utility maximizing behavior if and only if the symmetry and
second order conditions are satisfied.
The major advantage of this approach to modeling consumer behavior is that it
permits the specification of a system of Marshallian commodity demand equations
x D x.p; y/ that are consistent with utility maximization and with a very gen-
eral specification of the consumer’s utility function.2 In contrast, even if an accept-
2
The high degree of flexibility of V .p; y/ in representing utility functions (consumer preferences)
u.x/ follow from the fact that the first and second derivatives of V .p; y/ determine the first and
34 3 Static Utility Maximization and Expenditure Constraints

able measure of utility is defined and the first order conditions @u.x /=@xi
@.x  /=@xj
D pi =pj
.i; j D 1;


; N / are estimated, the corresponding Marshallian demand equations
x D x.p; y/ can be recovered explicitly only under very restrictive functional forms
for u.x/ (e.g. Cobb-Douglas).
A serious problem with this and all other consumer demand models derived from
microtheory (behavior of the individual consumer) is that the data is usually aggre-
gated over consumers. Difficulties raised by the use of such market data will be
discussed in a later lecture.

3.4 Profit maximization subject to budget constraints

The above model (3.1) of consumer behavior subject to a budget constraint is for-
mally equivalent to the following model of producer behavior subject to a budget
constraint
N
( )
X
max pf .x/ wi xi  .w; p; b/
x0
iD1
(3.16)
N
X
s.t. wi xi D b
iD1

since (3.16) and

max pf .x/  R.w; p; b/
x0
N
X (3.17)
s.t. wi xi D b
iD1

have identical solutions x assuming that the budget constraint is binding. Here
purchases of all inputs are assumed to draw upon the same total budget b, and there
is a single output with production function y D f .x/. In this case the solution x 
to (3.16) also solves the cost minimization problem

min wi xi  c.w; y  /
x0
(3.18)
s.t. f .x/ D y 

where y  D f .x  /. Note that cost minimization (conditional on y  ) is sufficient as

well as necessary for a solution to (3.16) (see pages 27–28), in contrast to the case
of profit maximization without a budget constraint.

second derivatives of the corresponding utility function u.x/ at x.p; y/. This can be seen by dif-
/ PN
ferentiating the first order conditions @u.x
@xi
D @V @y
.p;y/
pi , iD1 pi xi D y .i D 1;


; N /
for utility maximization and employing Roy’s Theorem, in a manner analogous to the derivation
of (1.6).
3.4 Profit maximization subject to budget constraints 35

The envelop relations and second order conditions for (3.16) and (3.17) are easily
derived as follows. (3.16) implies

.w; p; b; x/  .w; p; b/ fpf .x/ wxg  0

for all .w; p; b; x/ such that wx D b (3.19)
.w; p; b; x.w; p; b// D 0

or, substituting wx for b in .
/,

Q .w; p; x/  .w; p; wx/ fpf .x/ wxg  0 for all .w; p; x/

(3.19’)
Q .w; p; x.w; p; b// D 0

Thus Q w; p; x.w 0 ; p 0 ; b 0 / attains a minimum over .w; p/ at .w 0 ; p 0 /. This im-

plies the following first order conditions for a minimum:

@ Q .w 0 ; p 0 ; x 0 / @.w 0 ; p 0 ; b 0 /
 f .x 0 / D 0 i D 1;


; N
@p @p
(3.20)
@ Q .w 0 ; p 0 ; x 0 / @.w 0 ; p 0 ; b 0 / @.w 0 ; p 0 ; b 0 / 0
 C xi C xi0 D 0
@wi @wi @b
i.e. (3.16) implies

@.w; p; b/
y.w; p; b/ D
@p
(3.21)
@.w; p; b/=@wi
xi .w; p; b/ D i D 1;


; N
1 C @.w; p; b/=@b

Note that (3.21) reduces to Hotelling’s Lemma if @.w; p; b/=@b D 0 , i.e. if the
budget constraint is not
binding. The second order conditions for a minimum of
Q w; p; x.w 0 ; p 0 ; b 0 / over prices .w; p/ are Q w;p .w; p; x 0 / positive semidefi-


nite. Twice differentiating the identity Q .w; p; x/  .w; p; wx/ fpf .x/ wxg
with respect to prices .w; p/ yields

@2 Q .w; p; x/ @2 .w; p; b/ @2 .w; p; b/ @2 .w; p; b/ @2 .w; p; b/

 C xj C xi C xi xj
@wi @wj @wi @wj @wi @b @b@wj @b@b
@2 Q .w; p; x/ @2 .w; p; b/ @2 .w; p; b/
 C xi
@wi @p @wi @p @b@p
@2 Q .w; p; x/ @2 .w; p; b/ @2 .w; p; b/
 C xi
@p@wi @p@wi @p@b
@2 Q .w; p; x/ @2 .w; p; b/
 i; j D 1;


; N
@p@p @p@p
(3.22)
36 3 Static Utility Maximization and Expenditure Constraints

Thus profit maximization subject to a budget constraint wx D b implies that the

.N C 1/-dimensional matrix defined in (3.22) is symmetric positive semidefinite.
Likewise (3.17) implies

.w; p; x/  R.w; p; wx/ pf .x/  0 for all .w; p; x/

(3.23)
 .w; p; x.w; p; b// D 0

Proceeding as above we obtain the envelop relations

@R.w; p; b/
yD
@p
(3.24)
@R.w; p; b/=@wi
xi D
@R.w; p; b/=@b
and second order conditions analogous to (3.22) symmetric positive semidefinite.
Nevertheless, assuming an expenditure constraint over all inputs and a single
output, the simplest approach is to estimate a cost function c.w; y/ using Shephard’s
Lemma to obtain estimating equations

@c.w; y/
xi D i D 1;


; N: (3.25)
@wi
Under the above assumptions the hypothesis of cost minimization conditional on
y  exhausts the implications of the hypothesis of profit maximization subject to
a budget constraint, although of course this approach (3.25) still mis-specifies the
maximization problem (3.16) by treating output y as exogenous.
As a second and more interesting example of modeling budget constraints in
production, suppose that expenditures on only a subset of inputs are subject to a
budget constraint, (e.g. different inputs may be purchased at different times and
cash constraints may be binding only at certain times, or alternatively credit may be
available for the purchase of some but not all inputs). In this case the firm solves the
profit maximization problem
NA NB
( )
X X
max pf .xA ; xB / wiA xiA wiB xiB  .w; p; b/
xA ;xB 0
iD1 iD1
NAX
CNB
(3.26)
s.t. wiB xiB Db
i DNA C1

(3.26) implies

.w; p; x/  .w; p; wB xB / fpf .xA ; xB / wA xA wB xB g

0 for all .w; p; x/ (3.27)
D 0 for .w; p; x.w; p; b//
3.4 Profit maximization subject to budget constraints 37

Proceeding as before leads to the envelop relations

@.w; p; b/
yD
@p
@.w; p; b/
xiA D A
i D 1;


; N (3.28)
@wi
@.w; p; b/=@wiB
xiB D i D NA C 1;


; NA C NB
1 C @.w; p; b/=@b
and to the second order relations
@2 .w; p; x/ @2 .w; p; b/ @2 .w; p; b/ B @2 .w; p; b/ B @2 .w; p; b/ B B
 C xj C x i C xi xj
@wiB @wjB @wiB @wjB @wiB @b @b@wjB @b@b
i; j D 1;


; NB
2 2 2
@ .w; p; x/ @ .w; p; b/ @ .w; p; b/ B
 C xi i D 1


; NB j D 1;


; NA
@wiB @wjA @wiB @wjA @b@wjA
@2 .w; p; x/ @2 .w; p; b/ @2 .w; p; b/ B
B
 B
C xi i D 1


; NB j D 1;


; NA
@wi @p @wi @p @b@b
@2 .w; p; x/ @2 .w; p; b/
 i D 1;


; NA
@wiA @wjA @wiA @wjA
@2 .w; p; x/ @2 .w; p; b/
 i D 1;


; NA
@wiA @p @wiA @p
@2 .w; p; x/ @2 .w; p; b/

@p@p @p@p
(3.29)
The .NA C NB C 1/-dimensional matrix defined by (3.29) should be symmetric
positive semidefinite, assuming profit maximization subject to a binding budget con-
straint wB xB D b.
The above theory of profit maximization subject to a budget constraint can be
applied by first specifying a functional form .w; p; b/ for the profit function
.w; p; b/ and differentiating .w; p; b/ with respect to .w; p; b/ in order to obtain
the estimating equation

@ .w; p; b/
yD
@p
@ .w; p; b/
xiA D i D 1;


; NA (3.30)
@wiA
@ .w; p; b/=@wiB
xiB D i D NA C 1;


; NA C NB
1 C @ .w; p; b/=@b
38 3 Static Utility Maximization and Expenditure Constraints

where only inputs xB are subject to a budget constraint wB xB D b. The symmetry

2 2 @2 @2
conditions @w@i @wj D @w@j @wi , @w i @p
D @p@w i
, .i; j D 1;


; NA C NB / imply that
the output supply and factor demand equations (3.30) integrate up to a macrofunc-
tion .w; p; b/, and satisfaction of the second order conditions (3.29) (replacing
 with in the derivatives) implies that .w; p; b/ can be interpreted as a profit
function .w; p; b/ corresponding to the behavioral model (3.26). Note that models
such as (3.30) may be useful in testing whether expenditures on a particular input
draw on a binding budget, i.e. in testing whether the shadow price for the budget
constraint @ .w; p; b/=@b is significant in the demand equation for a particular in-
put.
For example, the functional form for .w; p; b/ could be hypothesized as
X X 
1 1 X 1 1
2 2 2
 Db aij wi wj C a0i p wi2 (3.31)
i j i

(a Generalized Leontief functional form with constant returns to scale). This leads
to the following functional form for the output supply and factor demand equations:
NAX
CNB   12
wi
yDb a0i
p
i D1
NAX
CNB   12
wj
xiA D b aij i D 1;


; NA
wi
j D1

1
b jND1
P A CNB
aij wj =wi 2
xiB D 1 1 1
PNA CNB PNA CNB 1
aj k wj2 wk2 C jND1
P A CNB
1C j D1 kD1
a0j p 2 wi2
i D NA C 1;


; NA C NB
(3.32)
Note that the particular functional form (3.31) implies @.w; p; b/=@b D .w; p; b/=b,
so that the demand equations for inputs xB can be simplified to
 NAX
CNB  12
b2
 
wj
xiB D aij i D NA C 1;


; NA C NB (3.33)
bC wi
j D1

However, in the absence of constant returns to scale, the demand equations for inputs
subject to a budget constraint generally will be nonlinear in the parameters to be
estimated.

References

1. Varian, H. R. (1992). Microeconomic Analysis, Third Edition. W. W. Norton & Company, 3rd
edition.
Chapter 4
Nonlinear Static Duality Theory (for a single
agent)

4.1 The primal-dual characterization of optimizing behavior

In previous lecture we have constructed primal-dual relations such as

G.w; p; x/  .w; p/ fpf .x/ wxg (page 33) (4.1a)

Q
G.p; y; x/  V .p; y/ u.x/ ((3.6) on page 30) (4.1b)
QQ
G.p; u; x/  E.p; u/ px (4.1c)

where the first term denotes the optimal value of profits/utility/expenditure for an
agent solving a profit maximization/utility maximization/cost minimization problem
and the second term denotes a feasible level of profits/utility/expenditure, respec-
tively. It is obvious that the primal-dual relations (4.1a)–(4.1b) attain a minimum
value (equal to zero) at an equilibrium combination of choice variables x and pa-
rameters: .w; p; x.w; p///.p; y; x.p; y//. Likewise the primal-dual relations c for
cost minimization attains a maximum value (equal to zero) at an equilibrium com-
bination of choice variables x and parameters: .p; u; x.p; u//. This implies that,
given equilibrium levels of x  of the agent’s choice variables x, the primal-dual re-
lations (4.1a)–(4.1b) attain a minimum over possible values of the parameters at the
particular level of the parameters for which x  solves the profit/utility maximization
problem:

given x 0  x.w 0 ; p 0 / solving a profit maximization problem (2.1):

.w 0 ; p 0 / min G.w; p; x 0 /  .w; p/ pf .x 0 / wx 0 D 0 (4.2)

˚
solves
w;p

given x 0  x.p 0 ; y 0 / solving a utility maximization problem (3.1):

39
40 4 Nonlinear Static Duality Theory (for a single agent)

.p 0 ; y 0 / solves Q
min G.p; y; x 0 /  V .p; y/ u.x 0 / D 0 s.t. px 0 D y
p;y
(4.3)

and similarly for case (4.1c), given x 0 D x.p 0 ; u0 / solving a cost minimization
problem (3.2):

p0 solves QQ
max G.p; u0 ; x 0 /  E.p; u0 / p x0 D 0 (4.4)
p

where u0  u.x 0 /. By analyzing the first order conditions for an interior solution
to these problems, we easily derived Hotelling’s Lemma (page 12), Roy’s Theorem
(page 29) and we can easily derive Shephard’s Lemma, respectively.
To repeat, it is obvious that the hypotheses of profit maximization, utility max-
imization and cost minimization imply that an equilibrium combination of choice
variables and parameters are obtained by solving (4.2)–(4.4), respectively. In this
sense a necessary condition for x 0 to solves a profit maximization problem (2.1)
conditional on prices .w 0 ; p 0 / is that .w 0 ; p 0 / solve (4.2), and similarly for utility
maximization and cost minimization.
A further question is: does x 0 solve a profit maximization problem (2.1) (e.g.)
conditional on .w 0 ; p 0 / if and only if (w 0 ; p 0 ) solves (4.2) conditional on x 0 ? The
answer is yes and this implies that problems (4.2)–(4.4) exhaust the implications of
behavioral models (2.1), (3.1) and (3.2), respectively, for local comparative static
analysis of changes in prices.
The intuitive explanation of this result is surprisingly simple (given the confusion
that has been raised in the recent past over this matter, especially Silberberg 1974,
pp. 159–72). For example, note that the profit maximizing derived demands x.w; p/
solving maxx fpf .x/ wxg  .w; p/ also solve

min G.w; p; x/  .w; p/ fpf .x/ wxg D 0 (4.5)

x

and note that any x such that G.w; p; x/ D 0 also solves maxx fpf .x/ wxg
conditional on .w; p/. Since the minimum value of G.w; p; x/ over .w; p/ is also 0,
it follows that any combination .w 0 ; p 0 ; x 0 / solving minw;p G.w; p; x 0 / also solves
minx G.w 0 ; p 0 ; x/, and conversely any .w 0 ; p 0 ; x 0 / solving minx G.w 0 ; p 0 ; x/ also
solves minw;p G.w; p; x 0 /.
Thus solving min˚w;p G.w; p; x 0 / is equivalent to solving the profit maximiza-
tion problems maxx p 0 f .x/ w 0 x . Also note that an (interior) global solution
.w 0 ; p 0 / to minw;p G.w; p; x 0 / implies for local comparative static purposes only
that
@G.w 0 ; p 0 ; x 0 / @G.w 0 ; p 0 ; x 0 /
D 0; D0 i D 1;


; N (4.6a)
@p @wi
 2
@ G.w 0 ; p 0 ; x 0 /

symmetric positive semidefinite (4.6b)
@p@w .N C1/.N C1/
4.2 Producer behavior 41

i.e. only the first and second order conditions for an interior minimum are relevant.1
The implication of the above argument is that (4.6) exhaust the restrictions placed
on the comparative static effects of (local) changes in prices .w; p/ by the hypoth-
esis of competitive profit maximization for the individual firm. Similar conclusions
hold for the cases of utility maximization (4.3) and cost minimization (4.4).
More generally, consider an optimization problem

max f .x; ˛/  .˛/

x
(4.7)
s.t. g.x; ˛/ D 0

where both the objective function f .x; ˛/ and constraint (or vector of constraint)
g.x; ˛/ D 0 are conditional on a vector of parameters ˛ (e.g. prices or parameters
shifting the price schedules facing the agent). Then x 0 solves (4.7) conditional on
˛ 0 if and only if ˛ 0 solves the following problem conditional on x 0 :

min G.x 0 ; ˛/  .˛/ f .x 0 ; ˛/

˛
(4.8)
s.t. g.x 0 ; ˛/ D 0

Therefore the first and second order conditions for a solution to problem (4.8) in
parameter (˛) space exhaust the implications of the behavioral model (4.7) for com-
parative static effects of (local) changes in parameters ˛.2

4.2 Producer behavior

Consider the following general profit maximization problem

max fR.x; ˛A / Q  ; ˛/  .˛; b/

c.x; ˛B /g  .x
x
(4.9)
s.t. c.x; ˛C / D b

where all prices may be endogenous to the producer and there is a budget constraint
(or vector of constraints) c.x; ˛C / D b limiting expenditures on at least some in-
puts. Hence R.x; ˛A / denotes total revenue as a function of the output level y (or
equivalent the inputs levels x, given a single output production function y D f .x/)
and parameters ˛A shifting the price schedules p.y; ˛A / facing the firm. c.x; ˛B /
denotes total costs as a function of the input levels x and the parameters ˛B shifting

1
The condition G.w 0 ; p 0 ; x 0 / D 0 for a global solution to minw;p G.w; p; x 0 / implies only
that .w 0 ; p 0 / D p 0 f .x 0 / w 0 x 0 , which is not directly relevant to local comparative statics.
2
Likewise, if we are only interested in the comparative static effects of changes in a sub-
0
set ˛B of parameters ˛, then x 0 solves (4.7) conditional on ˛ 0 if and only if ˛B solves
0 0 0 0
min˛B G.x ; ˛A ; ˛B / s.t. g.x ; ˛A ; ˛B / D 0. In turn the first and second order conditions
in the parameter ˛B subspace for this minimization problem exhaust the implications of (4.7) for
the comparative static effects of changes in parameters ˛B .
42 4 Nonlinear Static Duality Theory (for a single agent)

the factor supply schedules wi .x; ˛B / .i D 1;


; N / facing the firm. .˛; b/ de-
notes the dual profit function for (4.9), i.e. the relation between maximization attain-
able profits and parameters .˛; b/. Formally (4.9) allows for traditional monopoly/
monopsony behavior, the specification of financial constraints in terms of both fixed
cash constraints (at level b) and costs of borrowing the vary with the level of bor-
rowing (and the firm’s debt-equity ratio), the endogeneity of the opportunity cost
(value of forgone leisure) of farm family labor, etc.
Consider the corresponding minimization problem

min G.˛; b; x/  .˛; b/ fR.x; ˛A / c.x; ˛B /g

˛;b
(4.10)
s.t. c.x; ˛C / D b

or equivalently (substituting out the budget constraint c.x; ˛C / D b)

Q
min G.˛; x/   .˛; c.x; ˛C // fR.x; ˛A / c.x; ˛B /g (4.11)
˛

The analysis in the previous section implies that the following first and second or-
der conditions for an (interior) solution to (4.11) exhaust the implications of profit
maximization for the effects of local changes in parameters ˛: (using obvious vector
notation):

GQ ˛ .˛; x  / D 0
(4.12)
GQ ˛˛ .˛; x  / symmetric positive semidefinite
h i h i
Q Q Q ˛˛  @2 GQ
where x  D x.˛; b/ solves (4.9), and GQ ˛  @˛ @G
1



@G
@˛´ 1Z
, G @˛@˛ ZZ
.
Q
Since G.˛;  
x /   .˛; c.x ; ˛C // .x Q 
; ˛/, restrictions (4.12) can be rewritten
as

GQ ˛ .˛; x  /  ˛ .˛; b/ C b .˛; b/c˛ .x  ; ˛/ Q ˛ .x  ; ˛/ D 0 (4.13a)

GQ ˛˛ .˛; x  /  ˛˛ .˛; b/ C ˛b .˛; b/c˛ .x  ; ˛/ C b˛ .˛; b/c˛ .x  ; ˛/

C bb .˛; b/c˛ .x  ; ˛/c˛ .x  ; ˛/ C b .˛; b/c˛˛ .x  ; ˛/ (4.13b)

Q ˛˛ .x ; ˛/ symmetric positive semidefinite.
In order to derive a system of estimating equations for a behavioral model of the
type (4.9), we can begin by specifying a functional form .˛; b/ for the firm’s dual
profit function .˛; b/. Then we calculate corresponding envelop relations (4.13a).
Note, however, that calculation of these envelop relations also requires knowledge of
the derivatives c˛ .x; ˛/, R˛ .x; ˛/, C˛ .x; ˛/,3 i.e. we must specify functional forms
for R.x; ˛/, C.x; ˛/ and c.x; ˛/ in addition to the functional form for .˛; b/.
This does not appear to cause any serious problems: although these functions are
not independent of the functional form for .˛; b/, a flexible specification (see next
3
In the competitive case R.x; ˛A /  pf .x/, C.x; ˛B /  wx, c.x; ˛C /  wC xC ; So that
R˛A D y, C˛B D x, c˛C D xC .
4.3 Consumer behavior 43

lecture) of these functions and of .˛; b/ need not severely restrict the implicit
specification of the production technology y D f .x/.
After specifying the functional forms of .˛; b/ and R.x; ˛/, C.x; ˛/ and
c.x; ˛/, we then solve the envelope relations in the form

Q ˛ .x  ; ˛/ Q b .˛; b/ c˛ .x  ; ˛/ D ˛ .˛; b/ at b  c.x  ; ˛/ (4.14)

we see (using the Frobenius theorem) that these estimating equations integrate up
to a macrofunction  .˛; c.x  ; ˛// if and only if the symmetry condition (4.13b) is
satisfied. Thus we test the symmetry conditions GQ ˛˛ .˛; x  / ˛˛ .˛; b/ symmet-
ric and check the second order conditions GQ ˛˛ .˛; x  / positive semidefinite, using
(4.13b).

4.3 Consumer behavior

Consider the following general utility maximization problem

max u.x/  V .˛; y/

x
(4.15)
s.t. c.x; ˛/ D y

where these is a nonlinear budget constraint c.x; ˛/ D y, i.e. in general prices of

the commodities x can depend upon the levels of the commodities purchased by the
consumer as well as upon the parameters ˛. Also consider the corresponding cost
minimization problem
min c.x; ˛/  E.˛; u/
x
(4.16)
s.t. u.x/ D u
Comparative static effects for the maximization problem (4.15) can be analyzed
in term of the primal-dual relation

min G.˛; y; x/  V .˛; y/ u.x/

˛;y
(4.17)
s.t. c.x; ˛/ D y

or equivalently (substituting out the budget constraint)

Q
min G.˛; x/  V .˛; c.x; ˛// u.x/ (4.18)
˛

The first and second order conditions for a solution to (4.18) yield

GQ ˛ .˛; x  /  V˛ .˛; y/ C Vy .˛; y/ c˛ .x  ; ˛/ D 0 (4.19a)

44 4 Nonlinear Static Duality Theory (for a single agent)

GQ ˛˛ .˛; x  / V˛˛ .˛; y/ C V˛y .˛; y/ c˛ .x  ; ˛/ C Vy˛ .˛; y/ c˛ .x  ; ˛/

C Vyy .˛; y/ c˛ .x  ; ˛/ c˛ .x  ; ˛/ C Vy .˛; y/ c˛˛ .x  ; ˛/ (4.19b)
symmetric positive semidefinite
Equation (4.19a) c˛ .x  ; ˛/ D V˛ .˛; y/=Vy .˛; y/ are a generalization of Roy’s
Theorem.
Likewise comparative static effects for the minimization problem (4.16) can be
analyzed in terms of the primal-dual relation

max H.˛; x/  E.˛; u.x// c.x; ˛/ (4.20)

˛

The first and second order conditions for a solution to (4.20) yield

H˛ .˛; x  /  E˛ .˛; u / c˛ .x  ; ˛/ D 0 (4.21a)

H˛˛ .˛; x  /  E˛˛ .˛; u / c˛˛ .x  ; ˛/ symmetric negative semidefinite

(4.21b)
(4.21a) c˛ .x  ; ˛/ D E˛ .˛; u / is a generalization of Shephard’s Lemma.
In order to derive a generalization of the Slutsky equation, note that the Hick-
sian demands x h .˛; u/ solving the cost minimization problem (4.16) for param-
eters .˛; u/ also solve the utility maximization problem (4.15) for parameters
.˛; b/ D .˛; E.˛; u//, i.e.

xih .˛; u/ D xiM .˛; E.˛; u// i D 1;


; N for all ˛ (4.22)

Differentiating (4.22) with respect to ˛ (holding utility level u constant),

@xih .˛; u / @x M .˛; y/ @x M .˛; y/ @E.˛; u /

D i C i i D 1;


; N j D 1; 2
@˛j @˛j @y @˛j
(4.23)
Substituting the generalized Shephard’s Lemma (4.21a) into (4.23),

@xiM .˛; y/ @x h .˛; u / @xiM .˛; y/ @c.x  ; ˛/

D i i D 1;


N j D 1; 2
@˛j @˛j @y @˛j
(4.24)
where @c.x  ; ˛/=@˛j ¤ xj .˛; y/ except in the competitive case, where c.x; ˛/ 
PN
iD1 ˛i xi .˛i  wi /. Also note
h that thei second order conditions for cost minimiza-
h 
tion (4.21b) do not reduce to @x .˛;u
@˛
/
symmetric negative semidefinite except in
the competitive case.
An alternative generalization of the Slutsky equation that directly relates Mar-
shallian demands to the second order conditions for cost minimization (4.21b) can
be derived as follows. The first order conditions for cost minimization E˛ .˛; u /
c˛ .x  ; ˛/ D 0 (4.21a) can be rewritten as

E˛ .˛; u / D c˛ .x  ; ˛/ for all ˛ at x   x M .˛; E.˛; u // (4.25)

References 45

using (4.22). Differentiating (4.25) with respect to ˛,

h i
E˛˛ .˛; u / D c˛˛ .x  ; ˛/Cc˛x .x  ; ˛/ x˛M .˛; y/ C xyM .˛; y/E˛ .˛; u / (4.26)

substituting (4.25) into (4.26) and rearranging,

h i
c˛;x .x  ; ˛/ x˛M .˛; y/ C xyM .˛; y/ c˛ .x  ; ˛/ D E˛˛ .˛; u / c˛˛ .x  ; ˛/
symmetric negative semidefinite
(4.27)

by the second order conditions (4.21b) for cost minimization.

In order to derive a system of estimating equations for a behavioral model of the
type (4.15), we can begin by specifying a functional form for the consumer’s indirect
utility function V .˛; y/ and the budget constraint c.x; ˛/ D y. Then we calculate
corresponding envelope relations c˛ .x  ; ˛/ D V˛ .˛; y/=Vy .˛; y/ (4.19a) which
we then solve for consumer demand relations x  D x.˛; y/. Since utility maxi-
mization and cost minimization are equivalent, the Marshallian demand equations
integrate up to a macrofunction V .˛; y/ representing utility maximization behav-
ioral if and only if the corresponding envelope relations c˛ .x  ; ˛/ D E˛ .˛; u /
for cost minimization satisfy the symmetry and second order conditions (4.21b) for
integration up to a macrofunction E.˛; u / representing cost minimizing behavior.
These symmetry and second order conditions are related to the parameters of the
estimated Marshallian demand equations using the generalized Slutsky equations
(4.27).

References

1. Silberberg, E. (1974). A revision of comparative statics methodology in economics. Journal

of Economic Theory, pages 159–72
Chapter 5
Functional Forms for Static Optimizing Models

5.1 Difficulties with simple linear and log-linear Models

The main purpose of this lecture is to discuss the concept of flexible functional forms
and to present specific functional forms that are commonly employed with static
duality theory. In order to appreciate the potential value of such functional forms,
we begin with a discussion of the problem in using simpler linear and log-linear
models. For simplicity we restrict this discussion to cost-minimizing behavior.
First suppose that a simple linear model of a cost function c.w; y/ is formulated:
N
X
c D a0 C ai wi C ay y: (5.1)
iD1

If a producer minimizes his total cost of production, then Shephard’s Lemma applies
to the cost function. Applying Shephard’s Lemma to (5.1),

@c.w; y/
xi D D ai i D 1;


; N: (5.2)
@wi
i.e. the cost-minimizing factor demands x D x.w; y/ are in fact independent of
factor prices and the level of output. Alternatively suppose that factor demands are
estimated as a linear function of prices and output:
N
X
xi D ai 0 C aij wj C aiy y i D 1;


; N: (5.3)
j D1

However if factor demands are homogeneous of degree 0 in prices then (by Euler’s
theorem) jND1 @xi@w .w;y/
P
j
wj D 0 .i D 1;


; N:/ (see footnote 2 on page 3). So
(5.3) is consistent with cost minimizing behavior only if (5.3) reduces to

xi D ai 0 C aiy y i D 1;


; N: (5.4)

47
48 5 Functional Forms for Static Optimizing Models

i.e. cost minimizing factor demands are independent of factor prices (implying a
Leontief production function).
Of course these difficulties can be circumvented by first normalizing prices on a
numeraire price:
N  
X wj
xi D ai 0 C aij C aiy y i D 1;


; N: (5.5)
w1
j D2

Here the hypothesis that factor demands are homogenous of degree 0 in prices w is
imposed a priori by linearizing on normalized prices; so homogeneity cannot place
any further restrictions on the functional form (5.5) for factor demands. Thus, to the
extent that factor demands are homogenous of degree 0 in prices, (5.5) clearly is a
better specification of factor demands than is (5.3). Nevertheless, note that (5.5) does
impose an symmetry on the effects of the numeraire price w1 relative to the effects of
other prices on factor demands. Responses @c.w;y/
@wj
where (j ¤ 1) can be calculated
@c.w;y/ aij @c.w;y/
directly form (5.5) are simply @wj
D w1
, but responses @w
must be calcu-
@c.w;y/ PN1 @c.w;y/
lated indirectly from the homogeneity condition @w1
w1 C j D2 @w1 wj D 0
 
PN wj
as @c.w;y/
@w1
D j D2 a ij w1
.i D 1;


; N /.
Next consider simple log-linear models of cost minimizing behavior. A log-linear
cost function (corresponding to a cobb-Douglas technology)
N
X
ln c D a0 C ai ln wi C ay ln y (5.6)
i D1

implies that @@lnlnwci D ai , but @@lnlnwci  @c.w;y/

@wi
= wci D wicxi using Shephard’s
Lemma. Then the share of each input in that costs is independent of prices and
output:
wi xi
si  D ai i D 1;


; N: (5.7)
c
Alternatively consider log-linear factor demands:
N
X
ln xi D ai 0 C aij ln wj C aiy y i D 1;


; N: (5.8)
j D1

PN @xi .w;y/ 1 @ ln xi
Homogeneity implies j D1 @wj
= wj D 0 .i D 1;


; N / and @ ln wj
D
@xi .w;y/ xi
@wj
= wj using Shephard’s Lemma; so homogeneity of factor demands does not
imply the restriction jND1 aij D 0 .i D 1;


; N /. On the other hand, in the case
P
of log-linear consumer demands x D x.p; y/ conditional on prices p and income
y, the adding up constraint N @xi .p;y/
P
i D1 @y
pi D 1 (derived by differentiating the
budget constraint px D y with respect to y) is generally satisfied only if all income
elasticities are equal to 1 (see Deaton and Muellbauer 1980, pp. 16–17).
5.2 Second order flexible functional forms 49

5.2 Second order flexible functional forms

The previous section illustrated the severe restrictions implied by linea or log-linear
models of cost functions or (by extension) profit functions or indirect utility func-
tions. These functional forms imply extremely restrictive production functions and
behavioral relations. Likewise the most commonly employed production functions,
i.e. Cobb-Douglas or CES, place significant restrictions on behavioral relations (all
elasticities of substitution equal 1 or all elasticities of substitutions are equal, re-
spectively). Also note that a Cobb-Douglas production function is equivalent to a
Cobb-Douglas functional form for the associated cost function and profit function
(e.g. Varian 1984, p. 67).
The concept of second order flexible functional forms has been employed in or-
der to generate less restrictive functional forms for behavioral models (see Diewert
1971, pp. 481–507).
Suppose that the true production function, cost function, profit function or indi-
rect utility function for an agent is represented by the functional form f .x/. Then,
taking a second order Taylor series approximation of f .x/ about a point x0 ,
X @f .x0 / 1 X X @2 f .x0 /
f .x0 C x/  f .x0 / C xi C xi xj : (5.9)
i @xi 2 i j @xi @xj

Thus for a small x, f .x0 C x/ can be closely approximated  in terms of the 
@f .x0 / @2 f .x0 /
level of f at x0 (f .x0 /) and its first and second derivatives at x0 @x
; @x@x .
In turn, any other function g.x/ whose level at x0 can equal the level of f at x0
.g.x0 / D f .x0 // and whose first and second derivatives at x0 can equal  the first
@g.x0 / @f .x0 / @2 g.x0 / @2 f .x0 /
and second derivatives of f at x0 @x
D @x ; @x@x D @x@x can closely
approximate f .x/ for small variations in x about x0 .
To be more formal, g.x/ provides a second order (differential) approximation to
f .x/ at x0 if and only if

g.x0 / D f .x0 / (5.10a)

@g.x0 / @f .x0 /
D i D 1;


; N: (5.10b)
@xi @xi
@2 g.x0 / @2 f .x0 /
D i; j D 1;


; N: (5.10c)
@xi @xj @xi @xj

In general the true functional form f .x/ is unknown. Thus g.x/ provides a second
order flexible approximation to an arbitrary function f .x/ at x0 if conditions (5.10)
can be satisfied at x0 for any function f .x/. In other words, g.x/ is a second order
flexible functional form if, at any point x0 , any combination of level g.x0 / and
2 g.x /
derivatives @g.x
@x
0/
, @ @x@x0
can be attained. Thus a general second order flexible
functional form g.x/ must have at least 1 C N C N.N2C1/ free parameters (assuming
@2 g
@x@x
symmetric). If g.x/ is linear homogeneous in x, then (using Euler’s theorem)
50 5 Functional Forms for Static Optimizing Models

there are 1 C N restrictions.

N
X @g.x0 /
g.x0 / D xi 0
@xi
i D1
N
(5.11)
X @2 g.x0 /
0D xj 0 i D 1;


; N
@xi @xj
j D1

on g.x0 / and its first and second derivatives. Then a linear homogeneous second or-
der flexible functional form g.x/ has N.N2C1/ free parameters. Note that the number
of free parameters N.N2C1/ increase exponentially with the dimension N of x.
Consider a functional form c.w; y/ for a producer’s cost function.hc.w; y/ is ia
2
second order flexible functional form if c.w0 ; y0 /, @c.w@w
0 ;y0 / @c.w0 ;y0 /
, @y
, @ c.w 0 ;y0 /
@w@y .N C1/.N C1/
are not restricted a priori (except for homogeneity restrictions (5.11)) at .w0 ; y0 /.
Moreover a second order flexible approximation c.w; y/ to a true cost function im-
plies a second order flexible approximation to a true production function (see equa-
tion (1.5), (1.6) on page (1.5)).
Next consider a functional form .w; p/ for a firm’s profit function. .w; p/
is a second order flexible functional form if the combination .w0 ; p0 /, @.w@w0 ;p0 / ,
h 2 i
@.w0 ;p0 /
@p
, @ .w0 ;p0 /
@w@p
is not restricted a priori (except for homogene-
.N C1/.N C1/
ity restrictions). In addition a second order flexible approximation to a true profit
function implies a second order flexible approximation to a true production function
(see equation (2.5), (2.6) of page (2.5)). Likewise a second order approximation
V .p; y/ to a true indirect utility function implies a second order approximation to a
true utility function (structure of preferences).
Thus, to the extent that w, y is small over the data set .w; y/, a second or-
der flexible functional form for a producer’s cost function c.w; y/ provides a close
approximation to the true cost function and to the underlying production function.
Similar comments apply to profit functions and indirect utility functions. Note that
if prices w are highly co-linear over time then in effect the variation w in prices
may be small. Thus second order flexible functional forms may be useful in model-
ing behavior over data sets with very high multicollinearity. On the other hand, for
a cost function, changes in output .y/ are likely to be substantial over time and
not perfectly correlated with changes in factor prices .w/. In this case it may be
desirable to provide a third order or even higher order of approximation in output
y to the true cost function, e.g. by specifying a cost function c.w; y/ such that any
combination of the following cost and derivatives can be attained at a point .w0 ; y0 /
(subject to homogeneity restrictions):
5.3 Examples of second order flexible functional forms 51

c.w0 ; y0 / (5.12a)
@c.w0 ; y0 / @c.w0 ; y0 /
i D 1;


; N (5.12b)
@wi @y
2 2
@ c.w0 ; y0 / @ c.w0 ; y0 / @2 c.w0 ; y0 /
i; j D 1;


; N (5.12c)
@wi @wj @wi @y @y@y
@3 c.w0 ; y0 / @3 c.w0 ; y0 /
i D 1;


; N (5.12d)
@wi @y@y @y@y@y
Notice, however, that if third derivatives (5.12d) as well as (5.12a)–(5.12c) to be
free then a greater number of free parameters must be estimated in the model, which
implies a lost of degrees of freedom in the estimation.
In sum, there are two serious problems in the application of flexible functional
forms. First, the number of parameters to be estimated increases exponentially with
the number of variations (e.g. prices, outputs) included in the functional form and
with the order of the Taylor series approximation. Thus flexible functional forms
generally require a high level of aggregation of commodities; but consistent aggre-
gation of commodities is possible only under strong restrictions on the underlying
technology or preference structure (see next lecture). Second, when there is sub-
stantial variation in the data, the global properties of flexible functional forms (how
well these forms approximate the unknown true function f .x/ over large variation
in prices and output) become very important. Unfortunately the global properties of
many common flexible functional forms are not clear. It is often difficult to discern
whether these functional forms impose restrictions over large x that are unreason-
able on a priori grounds and hence seriously bias econometric estimates (see pages
232–236 of M. Fuss, D. McFadden, Y. Mundlak, “A Survey of Functional Forms
in the Economic Analysis of Production,” in M. Fuss, D. McFadden, Production
Economic: A Dual Approach to Theory and Applications, 1978 for a good early
discussion of this problem). Nevertheless the concept of flexible functional forms
appears to be useful in practice.

5.3 Examples of second order flexible functional forms

The most common flexible functional forms are the Translog and Generalized Leon-
tief, so we focus on these plus the Normalized Quadratic (which is the most obvious
flexible functional form). First consider dual profit functions .w; p/, which we
now write as .v/ where v  .w; p/.
The most obvious candidate for a flexible functional form for a dual profit func-
tion .v/ is the quadratic:
N N N
X 1 XX
.v/ D a0 C ai vi C aij vi vj (5.13)
2
iD1 iD1 j D1
52 5 Functional Forms for Static Optimizing Models

Note that this quadratic can be viewed as a second order expansion of  in pow-
ers of v. In the absence of homogeneity restrictions, (5.13) obviously is a sec-
ond order flexible functional form: differentiating twice yields @2 =@vi @vj D aij
.i; j D 1;


; N / i.e. each second derivative is determined as a free parameter aij ;
differentiating once yields @=@vi D ai C jND1 aij vj .i D 1;


; N / i.e. there
P
is a remaining free parameter ai to determine @=@vi at any level; and finally the
parameter a0 is free to determine  at any level. However linear homogeneity of
2
.v/ in v implies (by Euler’s theorem) jND1 @v@i @v
P
j
vj D 0 .i D 1;


; N / and in
PN
turn (by Hotelling’s Lemma) j D1 aij vj D 0. Thus the quadratic profit function
(5.13) reduces to a linear profit function:
N
X
 D a0 C ai vi (5.14)
i D1

which implies (using Hotelling’s Lemma) that output supplies and factor demands
are independent of prices v.
This problem is circumvented by defining the quadratic profit function in terms
of normalized prices:
N N N
X 1 XX
z .v/
 Q D a0 C ai vQ i C aij vQ i vQj (5.15)
2
i D1 i D1 j D1

where vQ i  vi =v0 .i D 1;


; N /, i.e. the inputs and outputs of the firm are indexed
i D 0;


; N and v0 is chosen as the numeraire. By construction (5.15) satisfies
the homogeneity condition (i.e. only relative prices vQ matter) so homogeneity does
not place any further restrictions on the functional form (5.15). Since vQ i  vi =v0
implies d vQ i D v10 dvi (for v0 fixed), the derivatives of  z .v/
Q and the correspond-
2 2
ing .v/ can be related simply as follows, @. z v/
Q
@vQi
D @.v/
@vi
, @@vQi.
z v/
Q
@vQj
@ .v/
D v10 @v i @vj
.i; j D 1;


; N /. The derivatives of .v/ with respect to v0 can then be recovered
from (5.15) using the homogeneity conditions (5.11). Often calculating the data vQ
from the data .v0 ; vi ;


; vN / we can assume without loss of generality (since only
relative prices matter) that v0  1 and specify the estimating equations as
N
@z
 .v/Q X
yi D D ai C aij vQj
@vQ i
j D1
(5.16)
N
@z Q
 .v/ X
xi D D ai aij vQj i D 1;


; N
@vQj
j D1

for outputs y and inputs x. The corresponding equation for the numeraire com-
modity can be recovered from (5.16) using homogeneity. These equations (5.16)
2 .
Q if @@v@
z .v/
integrate up to a macrofuction  Q v/
Q
Q vQ
is symmetric, i.e. if aij D aj i
5.3 Examples of second order flexible functional forms 53

@2 .
z v/
Q
 
.i; j D 1;


; N /, and profit maximization implies that the matrix @v@
Q vQ
D aij
is symmetric positive semidefinite.
Next consider a Generalized Leontief dual profit function:
N N N
X p 1 XX p p
.v/ D a0 C ai vi C aij vi vj (5.17)
2
iD1 iD1 j D1

p P p p
This is a second order expansion of  in powers of v. Note that j aij vi vj D
P p p P p p P p
 j aij vi vj whereas i ai vi D  i ai vi ; so .v/ D .v/ re-
quires a0 D 0, ai D 0 .i D 1;


; N /. Thus, imposing the restriction of linear
homogeneity, the Generalized Leontief .v/ can be rewritten as
N N
1 XX p p
.v/ D aij vi vj (5.18)
2
i D1 j D1

 1=2
@ P vj
Differentiating .v/ (5.18) once yields @v i
D aij C j ¤i aij vi
.i D
1;


; N /, and differentiating .v/ twice yields

@2 .v/ aij
Dp p j ¤i
@vi @vj vi vj
p (5.19)
@2 .v/ X vj
D aij 3=2 i; j D 1;


; N
@vi @vi vi
j ¤i

This provider a second order flexible form for .v/ subject to homogeneity restric-
tions. The corresponding estimating equations are
 1=2
@.v/ X vj
yi D D ai i C aij
@vi vi
j ¤i
 1=2 (5.20)
@.v/ X vj
xi D D ai i aij i D 1;


; N
@vi vi
j ¤i

for outputs y and inputs x. These equations integrate up to a macrofunction .v/ if

@2 .v/
@v@v
is symmetric, i.e. aij D aj i .i; j D 1;


; N / (see (5.19)). Profit maximiza-
2
tion implies that the matrix @@v@v
.v/
is symmetric positive semidefinite.
Third consider a Translog dual profit function
N N N
X 1 XX
ln .v/ D a0 C ai ln vi C aij ln vi ln vj (5.21)
2
iD1 iD1 j D1
54 5 Functional Forms for Static Optimizing Models

This is a second order expansion of ln  in powers of ln v. In order to determine

homogeneity restrictions, note that .v/ D .v/ implies ln .v/ D ln  C
ln .v/. Multiplying prices v by  in (5.21).
X X X
ln .v/ D a0 C ai ln.vi / C aij ln.vi / ln.vj /
i i j
X X X X
D a0 C ln  ai C ai ln vi C .ln /2 aij
i i i j
X X X X
C 2 ln  aij ln vi C aij ln vi ln vj
i j i j
X X X  X X
D ln .v/ C ln  ai C 2 ln  aij ln vi C .ln /2 aij
i i j i j
(5.22)

Thus the linear homogeneity condition ln .v/ D ln  C ln .v/ is satisfied if

N
X
ai D 1
iD1
N
(5.23)
X
aij D 0 i D 1;


; N
j D1

@ ln  @
Rearranging @ ln vi
D @vi
= vi as @
@vi
D @ ln 
@ ln vi


vi
and differentiating with respect to
vj ,

@2 .v/ @2 ln  @ ln vj  @ ln  @ 1
D
C

@vi @vj @ ln vi @ ln vj @vj vi @ ln vi @vj vi
 
 @ ln  @ ln 
D aij C i ¤j
vi vj @ ln vi @ ln vj
(5.24)
@2 .v/ @2 ln  @ ln vi  @ ln  @ 1 @ ln  
D
C

@vi @vi @ ln vi @ ln vi @vi vi @ ln vi @vi vi @ ln vi .vi /2
" #
@ ln  2
 
 @ ln 
D a i i C i; j D 1;


; N
.vi /2 @ ln vi @ ln vi

2 @ ln v
since @ ln@viln@ ln vj D aij (5.21), @vj j D v1j and @v @
j
D @@lnlnvj
vj . Thus, as in the
cost of the normalized Quadratic and Generalized Leontief, the differential equa-
tions yi .v/ D @.v/ @vi
(for outputs), xi .v/ D @.v/@vi
(for inputs) integrate up to a
macrofunction .v/ if aij D aj i .i; j D 1;


; N /. Profit maximization further
2 .v/
requires that the matrix @@v@v defined by (5.24) is symmetric positive semidefinite.
In contrast to the normalized Quadratic and Generalized Leontief, the Translog
model (5.21) is more easily estimated in terms of share equations rather than output
supply and factor demand equations per se. The elasticity formula @@lnlnvi D @v @ 
i
= vi
5.3 Examples of second order flexible functional forms 55

@ ln   @ ln  
and Hotelling’s Lemma yield yi D @ ln vi vi
, xi D @ ln vi vi
; so substituting for
@ ln 
and  from (5.21) yields estimating equations for y and x that are nonlinear in
@ ln vi
the parameters .a0 ;


; aN ; a11 ;


; aN N / which are to be estimated. On the other
hand, the elasticity formula and Hotelling’s Lemma directly imply piyi D @@lnlnvi
and wi yi D @@lnlnvi . Thus the Translog model (5.21) can be most easily estimated
in terms of equations for “profit shares”:
N
pi yi X
si  D ai C aij ln vj

j D1
(5.25)
N
wi xi X
si  D ai aij ln vj i D 1;


; N

j D1

The homogeneity condition (5.23) and reciprocity conditions aij D aj i .i; j D

2 .v/
1;


; N / can easily be tested, and the restriction @@v@v positive semidefinite can
be checked at data points .v; / using (5.24).
The only complication in this procedure for estimating Translog models arises
from the fact Pthat the sum
P of the output shares minus the sum of the input shares
pi yi w i xi
equals 1 : D  D 1. Writing these equations as si D
PN i  i 
˙.ai C j D1 aij ln vj / C ei .i D 1;


; N / where ei denotes the disturbance
for equation i, we see that this linear dependence between shares implies a linear
dependence between disturbances, i.e. ei for any equation must be a linear combi-
nation of the disturbances for all other equations. This in turn implies that one of
the share equations must be dropped from the econometric model for the purposes
of estimation, but this is not a serious problem since there are simple techniques
for insuring that econometric results are invariant with respect to which equation is
actually dropped.
The above functional forms are easily generalized to cost functions c.w; y/, and
the homogeneity, reciprocity and concavity conditions are calculated in an analo-
gous manner. The normalized quadratic cost function can be written as
N N N N
X 1 XX X
Q w;
c. Q y/ D a0 C ai wQ i Cay yC aij wQ i wQ j C aiy wQ i yCayy y 2 (5.26)
2
iD1 i D1 j D1 iD1

wi
where wQ i  w0
, and the corresponding factor demand equations are (using Shep-
hard’s Lemma)
Q w;
@c. Q y/
xi D
@wQ i
X N (5.27)
D ai C aij wQ j C aiy y i D 1;


; N
j D1
56 5 Functional Forms for Static Optimizing Models

The effects of changes in the numeraire price w0 can be calculated from (5.27) using
the homogeneity restrictions (5.11). Note that the parameters ay and ayy are not
Q w;
included in the factor demand equations, so that @c. Q y/=@y cannot be recovered
directly from (5.27). Nevertheless @c.w; y/=@y can be calculated indirectly from
(5.27) using the homogeneity condition @c.w; y/=@y D @c.w; y/=@y and the
reciprocity relations @2 c.w; y/=@wi @y D @2 c.w; y/=@y@wi .i D 1;


; N / plus
Shephard’s Lemma (See Footnote 2 on page 19). Alternatively the cost function
(5.26) can be estimated directly along with N 1 of the factor demand equations
(5.27).
The Generalized Leontief cost function is often written as
N N N
1 XX p p X
c.w; y/ D y aij wi wj C y 2 aiy wi (5.28)
2
i D1 j D1 i D1

where aiy D 0 i D 1;


; N / implies that the production function is constant

returns to scale. The corresponding factor demand equations are

@c.w; y/
xi D
@wi
  12 (5.29)
X wj
D ai i y C aij y C aiy y 2 i D 1;


; N
wi
j ¤i

The Translog cost function is

N N N
X 1 XX
ln c.w; y/ D a0 C ai ln wi C ay ln y C aij ln wi ln wj
2
i D1 i D1 j D1
(5.30)
N
X
2
C aiy ln wi ln y C ayy .ln y/
i D1

and the corresponding factor share equations are

N
wi xi X
si  D ai C aij ln wj C aiy ln y i D 1;


; N (5.31)
c
j D1

As in the case of the Translog profit function, N

P
i D1 si D 1, so that one share
equation must be deleted from the estimation.
Finally, we briefly consider indirect utility functions V .p; y/ where p denotes
prices for consumer goods and y now denotes consumer income. The Normalized
Quadratic indirect utility function can be written as
N N N
X 1 XX
Vz .p/
Q D a0 C ai pQi C aij pQi pQj (5.32)
2
iD1 i D1 j D1
5.4 Almost ideal demand system (AIDS) 57

where pQi  pi =y, so that (5.32) is by construction homogeneity of degree 0 in

.p; y/. @V .p; y/=@y can be calculated from (5.32) using the condition N @V .p;y/
P
i D1 @pi
pi C
@V .p;y/
@y
y D 0 implied by zero homogeneity. Then the consumer demand equations
can be calculated using Roy’s identity xi D @V@p .p;y/ @V .p;y/
i
= @y .i D 1;


; N /.
Roy’s identity generally leads to equations that are nonlinear in the parameters (a)
to be estimated. A simple example of a Translog indirect utility function is
N N N
X 1 XX
Q D a0 C
ln V .p/ ai ln pQi C aij ln pQi ln pQj (5.33)
2
i D1 iD1 j D1

where again pQi  pi =y .i D 1;


; N /. (see Varian 1984, pp. 184–186 for further
examples of functional forms for indirect utility functions).

5.4 Almost ideal demand system (AIDS)

Suppose consumer expenditure function in the form

log E.p; u/ D a.p/ C b.p/ u p D .p1 ;


; pM / (5.34)

where
X 1X X 
a.p/ D ˛0 C ˛i log pi C r log pi log pj
i 2 i j ij
Y (5.35)
b.p/ D ˇ0 pi ˇ i ( ˇ0 p1 ˇ1


pM ˇM
i

Since E.p; u/ D E.p; u/, the following relation apply,

M
X
˛i D 1
i D1
M M M
(5.36)
X X X
rij D rij D ˇi D 0
i D1 j D1 i D1

By Shephard’s Lemma.

@ log E @E pi xi pi

D  si (cost share i ) (5.37)
@ log pi @pi E E
So that cost share equations are attained from Shephard’s Lemma by differentiating
(5.34)–(5.35):
58 5 Functional Forms for Static Optimizing Models
@ log E
si D
@ log pi
X @b (5.38)
D ˛i C r  log pj C u

j ij „ƒ‚… @ log pi
„ ƒ‚ … log E a.
/
@a b.
/ by (??)
@ log pi

This (5.38) simplifies to

N
X
si D ˛i C rij log pi C ˇi log Y =P (5.39)
i D1

where Y is consumer expenditure .E/ and P is a price index given by

M M M
X 1 XX
log P D ˛0 C log pi C rij log pi log pj (5.40)
2
j D1 iD1 j D1

and
1 
rij  .r C rji / for all i; j (5.41)
2 ij
Except for the price index P , demands (5.39) are linear is coefficients. Homogeneity
and symmetric imply.
M
X
rij D 0 i D 1;


; M (homogeneity) (5.42a)
j D1
rij D rj i for all i; j D 1;


; M: (5.42b)

In practice P is usually approximated by an appropriate arbitrary pure index, e.g.

M
X
log P  sj log pj (5.43)
j D1

and then (5.39) is estimated.

5.5 Functional forms for short-run cost functions

c.w; y; K/ CRTS f .x; K/ D c.x; K/ ) c.w; y; K/ D c.w; y; K/

The issue: form for c.w; y; K/ should be 2nd-order flexible functional form with
and without CRTS.
References 59

5.5.1 Normalized quadratic: c  D c=w0 , w D w=w0 .

e.g.
 

X 1X X  X 
c D a0 C C ai wi  
aij wi wj y C b0 C bi wi y 2
i 2 i j i
 X   X   X p
C c0 C ci wi K C d0 C di wi K 2 C e0 C ei wi Ky
i i i

) (under CRTS)
 

X
 1X X  
 X   X p
c D a0 C ai wi C aij wi wj yC c0 C ci wi KC e0 C ei wi Ky
i 2 i j i i

5.5.2 Generalized Leontief:

 X X
p p

1 X  X  X 
cD aij wi wj y C bi wi y 2 C c i wi K C di wi K 2
2 i j i i i
X p
C ei wi Ky
i

) (under CRTS)
 X X
p p

1 X  X p
cD aij wi wj yC ci wi KC ei wi Ky
2 i j i i

5.5.3 Translog:

X 1X X
log c D a0 C ai .log wi / C aij .log wi /.log wj / C b0 log y C b1 .log y/2
i 2 i j
X X
C bi .log wi /.log y/ C c0 log K C c1 .log K/2 C ci .log wi /.log K/ C e.log y/.log K/
i i
) (under CRTS)

log c.w; y; K/ D log  C log c.w; y; K/

References

1. Deaton and Muellbauer (1980). Economics and Consumer Behavior. pp. 16–17
60 5 Functional Forms for Static Optimizing Models

2. Diewert W. (1971), An Application of the Shephard Duality Theorem: A Generalized Leontief

Production Function, Journal of Political Economy, 1971, pp. 481–507
3. Fuss M., McFadden D. (1978). Production Economic: A Dual Approach to Theory and Ap-
plications
4. Varian H. (1984), Microeconomic Analysis, second edition,WW Norton & Co. pp. 67
Chapter 6
Aggregation Across Agents in Static Models

6.1 General properties of market demand functions

In previous lecture we characterized the behavior of an individual producer or con-

sumer at a static equilibrium. However in practice we often have data aggregated
over agents rather than data for the individual producers or consumers. This leads to
the following question: do the behavioral restrictions that apply to data at the firm
or consumer level also apply to data that has been aggregated over agents?
In several simple behavioral models the answer to this question is “yes”. The
most obvious example is the simple theory of static profit maximization, where all
input are free variable and all firms face the same prices w, p. Then the market
prices w, p correspond exactly to the parameter facing the individual firms so that
industry output supply and factor demand relations Y D Y .w; p/, X D X.w; p/,
where Y D fFD1 y f and X D f x f , do not misrepresent the parameters facing
P P
individual firms. More precisely,
N
( )
f
X
 f .w; p/ py f wi xi  0 for all .w; p/ f D 1;


; F
i D1
N
( )
wi xif
X X X
)  f .w; p/ py f  0 for all .w; p/ (6.1)
f f i D1
N
( )
X
i.e. ….w; p/ pY wi Xi  0 for all .w; p/
iD1

By (6.1), we can develop the properties of the industry profit function ….w; p/
and industry output supplies Y D Y .w; p/ and derives demands X D X.w; p/

61
62 6 Aggregation Across Agents in Static Models

in essentially the same manners as in the case of data for the individual firm (see
section 4.1 of lecture 4 on nonlinear duality).1
As a second example consider the case where each consumer maximizes utility
subject to the value i pi wif of his initial endowments w f D .w1f ;


; wN
f
P
/ of
f
the N commodities rather than an income y that is independent of commodity
prices p, i.e. each firm f solves

max uf .x f /
xf
N N (6.2)
pi xif D pi wif :
X X
s.t.
iD1 iD1


Then the solution x f to (6.2) is conditional on .p; w f / and in turn aggre-

gate market demands f x f are conditional on .p; w 1 ;


; w F /. If the en-
P

dowments w f of consumer f are constant over time for each consumer f D


1;


; F , then for purposes of estimation the market demands f x f are in
P
effect conditional only on the prices p which are common to each consumer:

x  D x  .p/ D f x f .p/. Then the aggregate market demand relation x D x.p/
P

are well defined and inherit all linear restrictions on x f .p/. This includes the
homogeneity restrictions x f .p; y f / D x f .p; y f /, which can be expressed as
x f .p; w f / D f f
h x .p; w /, but apparently excludes the nonlineari second or-
f f f
i h H
xi y f p; y xj  xi pj p; uf  symmetric
f



der conditions xi pj p; y
negative semidefinite.
In general the answer to the above question is “no”, i.e. the behavioral restric-
tions that apply to data at the level of the individual agent generally do not apply to
data that has been aggregated over agents. This question has been addressed in the
context of utility maximization by consumers:

max uf .x f /
xf

N ! x f D x f .p; y f /
pi xif
X
f
s.t. Dy
i D1

which implies the Slutsky relations

xpf .p; y f / C xyff .p; y f /x f .p; y f / symmetric negative semidefinite (6.3)

(property (3.4).b on page 31). It has been shown that, in the absence of special
restrictions on utility functions uf .x f / or on the distribution of income y f over
consumers, the above restrictions (6.3) on demands of the individual P consumer do
not carry over to aggregate demands X D X.p; Y / where X  f x f and Y 

1
The one exception concerns whether these relations .w; p/, y.w; p/, x.w; p/ are well de-
fined in cases of free entry and exit to the industry (see footnote 2 on page 19).
6.1 General properties of market demand functions 63

y f . Indeed, if the number of consumers is equal to or greater than the number

P
f
of goods N , then any continuous function X.p; Y / that satisfies Walras’ Law
Walras Law is a principle in general equilibrium theory asserting that when con-
sidering any particular market, if all other markets in an economy are in equilib-
rium, then that specific market must also be in equilibrium. Walras Law hinges on
the mathematical notion that excess market P demandsP (or, conversely, excess mar-
ket supplies) must sum to zero. That is, XD D XS D 0. Walras’ Law is
named for the mathematically inclined economist Leon Walras, who taught at the
University of Lausanne, although the concept was expressed earlier but in a less
mathematically rigorous fashion by John Stuart Mill in his Essays on Some Unset-
tled Questions of Political Economy (1844). (in particular the adding up properties
PN @xi .p;y/
pi D 1, N @xi .p;y/
pi C @xi@y
.p;y/
P
iD1 @y iD1 @pj
y D 0, j D 1;


; N in the
differentiable case) can be generated by some set of utility maximizing consumers
with some distribution of income (see Debrew 1974, pp. 15–22; Sonnenschein 1973,
pp. 345–354).
There is a relatively simple way of addressing the above question. Following
Diewect (1977, page 353–362) write the Slutsky relations (6.3) in the form

xpf .p; y f / D K f xyff .p; y f / x f .p; y f / (6.4)

N N 1N
N N N 1

where K f  xphf .p; uf  / is the hHickscian substitution matrix

i which is symmetric
negative semidefinite. In general xyff .p; y f /x f .p; y f / is neither symmetric nor
negative semidefinite. Choose N 1 linearly independent vector v1 ; v2 ;


; vN 1
such that x f vn D 0 for n D 1;


; N 1, and write v1 ;


; vN 1 in matrix
1N N 1
form as V . Pre and post-multiplying (6.4) by V ,
N .N 1/

VT xpf V D V T Kf V V T xyf x f V D V T K f V (6.5)

.N 1/N
N N N .N 1/

since x f V D 0. Since K f  xphf .p; uf  / is symmetric negative semidefinite,

T
(6.5) implies that V f xpf .p; y f /V f is symmetric negative semidefinite.
Now sum (6.4) over consumers f D P 1;


; F to obtain the matrix of partial
derivatives of aggregate demands X D f x f with respect to prices p as

xyff .p; y f /x f .p; y f /:

X X
Xp .p; y f / D Kf (6.6)
f f

Let vQ 1 ;


; vQ N F be N F linearly independent vectors, each of which is orthog-

onal to x 1 ;


; x F , the set of initial demand vectors of the F consumers. Define the
N  .N M / matrix VQ D ŒvQ 1 ;


; vQ N M . Pre and postmultiplying (6.6) by VQ ,

K f xyff x f VQ D VQ T
X X X
VQ T Xp VQ D VQ T K f VQ VQ T K f VQ (6.7)
.N M /.N M / f f f
64 6 Aggregation Across Agents in Static Models

since VQ T x f D 0N F for all f D 1;


; F . Therefore K f symmetric negative

semidefinite (f D 1;


; F ) implies

VQ T Xp VQ is a symmetric negative semidefinite matrix of dimension.N F /.N F /:

(6.8)
(6.8) can be viewed as restrictions places on aggregate consumer demands
X.p; y 1 ;


; y F / by symmetry negative semidefiniteness of Hicksian demand re-
sponses xphf .p; uf  / .f D 1;


; F /. Moreover these are essentially the only re-
strictions on the first derivatives of market demand functions (aside from adding
up constraints) (Mantel 1977). Note that if the number of consumers F is equal to
the number of goods N , then (6.8) places 0 restrictions on aggregate demands (the
dimensions of the matrix of restrictions are .N F /  .N F / equals 0  0 in
this case). This is consistent with the result of Debreu and Sonnenschein that sym-
metry negative semidefiniteness of Hicksian demand responses xphf .p; uf  /  K f
places no restrictions on aggregate demands when the number of consumers is equal
to or greater than number of goods.
In sum, in general (i.e. in the absence of restrictions on production/utility func-
tions or on the distribution of exogenous parameters across agents) the microthe-
ory of the individual agent cannot be applied to data that has been aggregated over
agents. In the next two sections we investigate how restrictions on production/utility
functions and on the distribution of parameters across agents can influence this con-
clusion.

6.2 Condition for exact linear aggregation over agents

Here we consider restrictions on production and utility functions that imply consis-
tent aggregation over agents for any distribution of the exogenous parameters that
vary over agents. Fist, consider the conditional factor demands
 
xif D xif w; y f i D 1;


; N f D 1;


; F (6.9)

where w  vector of factor prices (which are common

P f to each firm) P and yf 
f
output level of firm f . Aggregate factor demands f x D X.w; f y / exist if
and only if
 P  P
Xi w; f y f D f xif w; y f


i D 1;


; N (6.10)

for all data .w; y 1 ;


; y F /. Differentiating (6.10) w.r.t. y f ,

6.2 Condition for exact linear aggregation over agents 65

@Xi .w; y/ @y @xif .w; y f /

D
@y @y f @y f
(6.11)
@Xi .w; y/ @x f .w; y f /
i.e. D i f i D 1;


; N f D 1;


; F
@y @y

since @y=@y f D 1 in the case of linear aggregation y D f y f .

P

P on the distribution of total output y over firms,

Thus, in the absence of restrictions
aggregate factor demands X.w; f y f / exist if and only if @x f .w; y f /=@y f are
constant across all firms (for a given w). This condition implies that conditional
factor demand equations for firms are of the following form:

xif D ˛i .w/y f C ˇif .w/ i D 1;


; N f D 1;


; F (6.12)

where the function ˛i .w/ is invariant over firms. (6.12) is called a “Gorman Polar
Form”
Gorman polar form is a functional form for indirect utility functions in eco-
nomics. Imposing this form on utility allows the researcher to treat a society of
utility-maximizers as if it consisted of a single individual. W. M. Gorman showed
that having the function take Gorman polar form is both a necessary and sufficient
for this condition to hold.. Here the scale effects @x f .w; y f /=@y f are independent
of the level of output, which implies that the production function y f D y f .x f / is
“quasi-homothetic”:

Fig. 6.1 quasi-homothetic

production function xf1

y 1f
0f
y
xf2
0

Here, as output y f expands and factor price w remain constant, the cost minimizing
level of inputs increase along a ray (straight line) in input space. Condition (6.12)
implies that aggregate demands have a Gorman Polar Form2 :
2 4x f .w;y f /
The more restrictive assumption of hornatheticity implies that the expansion path 4y f
is
f
a ray through the origin. Given the Gorman Polar Form (6.12), homotheticity requires ˇ .w/ D 0
(which in turn implies the stronger assumption of constant restrun to scale).
66 6 Aggregation Across Agents in Static Models

xif .w; y f / D ˛i .w/ ˇif .w/ D ˛i .w/Y C ˇi .w/

X X X
Xi D yf C
f f f (6.13)
i D 1;


; N:

Condition (6.12) implies the existence of aggregate conditional factor demands

satisfying (6.10), irrespective of whether producers are minimizing cost. The next
question is: under what conditions do the restrictions on cost minimizing P factor
demands at the firm level generalize to aggregate factor demands X.w; f y f /?
The simplest way to answer this question is as follow: a Gorman Polar Form for
aggregate demands (6.13) implies that identical demand Prelations could be gener-
ated by a single producer with an output level equal to f y f , and by assumption
this producer is a cost minimizer. Therefore aggregate demands of a Gorman Po-
lar Form necessarily inherit the properties of cost minimizing factor demand for a
single agent.
To answer the above question more rigorously, first note that the assumption of
cost minimization and condition (6.12) for existence of aggregate factor demands
X.w; f y f / are jointly equivalent to the assumption of cost minimization plus a
P
Gorman Polar Form for the cost function of firm f :

c f .w; y f / D a.w/y f C b f .w/; f D 1;


; F: (6.14)

@c f .w;y f /
Sufficiency is obvious: (6.14) and Shephard’s Lemma imply xif .w; y f / D @wi
D
@a.w/ f @b f .w/ @a.w/ @b f .w/ f
@wi
y C @wi
where @wi
D ˛.w/, D ˇ .w/, and necessity follows
@wi
simply by integrating (6.12) up to a cost function using Shephard’s Lemma. This
implies the existence of an industry cost function which is also a Gorman Polar
Form:
X X  
c f .w; y f / D a.w/y f C b f .w/
f f
X X
D a.w/ yf C b f .w/
f f
X (6.15)
D a.w/ y f C b.w/
f
 X 
D C w; yf
f

Moreover this industry cost function C.w; f y f / D a.w/ f y f C b.w/

P P
inherits all the properties of cost minimization by a firm. To see this, note that
6.2 Condition for exact linear aggregation over agents 67
  N
wi xif  0
X
c f w; y f .x f / for all w f D 1;


; F
iD1
  N
wi xif  0
X XX
) c f w; y f .x f / for all w
f f i D1
N
wif xif  0
X X XX
, a.w/ y f .x f / C b f .w/ for all n (6.16)
f f f iD1

X N
X
, a.w/ y f C b.w/ wi xi  0 for all w
f iD1

 X  N
X
, C w; yf wi xi  0 for all w
f
i D1

and likewise
  N
wi xif D 0
X
c f w; y f .x f / f D 1;


; F
i D1
(6.17)
 X  N
X
) C w; yf wi xi D 0
f
iD1

PN
yf /
P
Thus the aggregate primal-dual relation C.w; wi xi has the same
f i D1
PN f f f
properties as the cost minimizing firm’s primal-dual c .w; y /
P f iD1 wi xi .f D
1;


; F /, and in turn C.w; f y / has the same properties as the cost minimizing
c f .w; y f /:

Property 6.1.
a) C.w; f y f / is increasing in w, f y f ;
P P
 
b) C.w; f y f / D C w; f y f ;
P P

c) C.w; f y f / is concave in w;
P

d) C.w; f y f / satisfies Shephard’s Lemma3

P

 P 
 X  @C w; f y f
Xi w; yf D i D 1;


; N
f @wi

3
Conditions Prop.6.1.a–c are satisfied for C.w; f y f /  a.w/ f y f Cb.w/ if a.w/ > 0
P P
and both a.w/ and b.w/ are increasing, linear homogeneous and concave in w.
68 6 Aggregation Across Agents in Static Models

In turn the aggregate demands satisfy the restrictions as the cost minimizing de-
mands for a firm:

Property 6.2.
 X   X 
a) X w; y f D X w; yf
" Pf # f
@X.w; f y f /
b) is symmetric negative semidefinite.
@w
N N

Similar results hold for linear aggregation of consumer demand equations xif D
xif.p; y f / where p price of consumer goods (which are assumed to be identical
for all consumers) and y f  exogenous income of consumer f . The conditions for
existence of aggregate demands are
 X  X  
Xi p; yf D xif p; y f i D 1;


; N (6.18)
f f

and the conditions are equivalent to Gorman Polar Forms

 
xif p; y f D ˛i .p/y f C ˇif .p/ i D 1;


; N f D 1;


; F (6.19)

which imply
 X  X
Xi p; y f D ˛i .p/ y f C ˇi .p/ i D 1;


; N: (6.20)
f f

In order to determine rigorously the restrictions that permit aggregate demands

(6.20) to inherit the properties of utility maximization, first note that utility max-
imization by a consumer is essentially equivalent to cost minimization (see page
28):
N
pi xif D E f .p; uf /
X
min
xf (6.21)
iD1

s.t. uf .x f / D uf

where uf is the maximum attainable utility level for the consumer given his budget
constraint px f D y f . An aggregate cost function E.p; f uf / exists if and only
P
if  X  X
E p; uf D E f .p; uf / (6.22)
f f

and this condition is equivalent to the existence of Gorman Polar Forms

E f .p; uf / D a.p/uf C b f .p/ f D 1;


; F (6.23)

6.2 Condition for exact linear aggregation over agents 69

which in turn implies

 X  X
E p; uf D a.p/ uf C b.p/ (6.24)
f f

(6.23) and (6.24) imply

  N
pi xif  0
X
E f p; uf .x f / for all p f D 1;


; F
i D1
 X  N
X
) E p; uf pi xi  0 for all p
f
iD1
(6.25)
  N
pi xif
X
f f f
E p; u .x / D0 f D 1;


; F
i D1
 X  N
X
) E p; uf pi xi D 0:
f
i D1

Therefore E.p; f uf / inherit the properties of cost minimization, and Shephard’s

P
Lemma implies that
 P 
 X  @E p; f uf
Xih p; uf D
f @pi (6.26)
@a.p/ X f @b.p/
D u C i D 1;


; N
@pi f @pi
In order to relate (6.26) to Marshallian demands, note that (6.23) and utility maxi-
mization by consumers f imply

y f D a.p/uf C b f .p/
 yf b f .p/ (6.27)
) uf D f D 1;


; F
a.p/
i.e. the consumer’s inherit utility function has the Gorman Polar Form

V .p; y f / D a.p/y
Q f
C bQ f .p/ (6.28)

Q
where a.p/ Q
D 1=a.p/, b.p/ D b f .p/=a.p/. Substituting (6.27) into (6.26),
70 6 Aggregation Across Agents in Static Models

@a.p/ X y f b f .p/
Xi D
@pi a.p/
f

@a.p/ 1 X f @a.p/ X b f .p/ @b.p/

D y C i D 1;


; N
@pi a.p/ @pi a.p/ @pi
f f
(6.29)

(6.29) implies the existence of aggregate Marshallian demands of a Gorman Polar

Form:
QQ
 X  X
Xi p; QQ
y f D a.p/ y f C b.p/ i D 1;


; N (6.30)
f f

where a.p/ D @a.p/ 1

@pi a.p/
, b.p/ D @a.p/ b.p/
@pi a.p/
C @b.p/
@pi
.
Thus cost minimization by individual consumers and a Gorman Polar Form
(6.18/6.30) that satisfy restrictions analogous to restriction on Marshallian demands
by individual utility maximizing consumers. The restriction (6.1) implied by cost
minimization are satisfies for E.p; f uf / (6.24) if
P

a.p/ > 0I (6.31a)

a.p/, b.p/ increasing and linear homogeneous in P ; (6.31b)

@2 a.p/
   2 
@ b.p/
, symmetric, negative semidefinite. (6.31c)
@p@p @p@p
These restrictions can be tested or imposed on the Gorman Polar Form X.p; f y f /
P
(6.30)
Alternatively we could try P to incorporate the utility maximization restrictions
into aggregate demands X.p; f y f / directly by specifying consumers’ indirect
utility functions V f .p; y f / as Gorman Polar Form:

Q
QQQ
V f .p; y f / D a.p/y f
C bQQ f .p/: (6.32)

However inverting this relation yields

 QQQ
f uf b.p/
y D (6.33)
QQQ
a.p/
i.e.
Q
1 bQQ f .p/
E f .p; uf / D uf f D 1;


; F: (6.34)
QQQ
a.p/ QQQ
a.p/
Thus an indirect utility function V f .p; y f / has a Gorman Polar Form if and only
if the corresponding cost function E f .p; uf / has a Gorman Polar Form. Therefore
the above analysis in term of cost minimizing behavior exhaust the restrictions per-
6.2 Condition for exact linear aggregation over agents 71

mitting the existence of aggregate Marshallian demands X.p; f y f / that satisfy

P

the same restrictions as Marshallian demands x f .p; y f / for an individual utility

maximizing consumer.
Note that these restrictions (6.23–6.24, 6.30–6.31) on preference structure do
imply mild restriction on the distribution of expenditures over consumers:

E f D a.p/uf C b f .p/ a.p/ > 0

f f
(6.35)
) y > b .p/ f D 1;


; F

Aside from this restriction, Gorman Polar Forms permit aggregate demands to in-
herit the properties of utility maximizing demands irrespective of the distribution of
expenditure over consumers. b f .p/ can be viewed as the agent’s “committed ex-
penditure” at prices p (since it is independent of utility level uf ), and y f b f .p/
can be defined as the corresponding “uncommitted expenditure”.
Aggregation problems also arise when there are variations in prices over agents,
and these problems can be severe. For example, suppose that profit maximizing
competitive firms in an industry are distributed across different regions of the coun-
try and as a result
P firms face different output Pprices p f . Aggregate
P demands can
be defined as f x D X.w; f f p /, f y f D Y .w; f f p f / where
f f
P
f
P
f f p denotes a (weighted) average output price p. N Aggregate demands and
supplies exist if
 X  X
Xi w; f p f D x f .w; p f / i D 1;


; N
f f i
 X  X (6.36)
Y w; f p f D y f .w; p f /
f f

These conditions are satisfied if the aggregate demands and supplies have Gorman
Polar Forms:
X
Xi D ˛i .w/ f p f C ˇi .w/ i D 1;


; N
f
X (6.37)
Y D ˛0 .w/ f p f C ˇ0 .w/
f

Now suppose further that these aggregate relation are to inherit the properties
of profit maximizing demand and supply relations for the individual firm. In the
absence of any restrictions on the distribution of prices p f over firms, this requires
the firm and industry profit functions to have the following Gorman Polar Forms

 f .w; p f / D a.w/p f C b f .w/ f D 1;


; F

 X
f
 X
f
(6.38)
… w; f p D a.w/ f p C b.w/
f f

However Hotelling’s Lemma now implies that output supplies are independent of
output prices:
72 6 Aggregation Across Agents in Static Models

@ f .w; p f /
y f .w; p f / D D a.w/ f D 1;


; F (6.39)
@p f
using (6.38).
In sum, Gorman Polar Forms (GPF) are both necessary and sufficient for ex-
act linear aggregation over agents. Unfortunately these functional forms are fairly
restrictive. For example, GPF conditional factor demands x D x.w; y/ imply lin-
ear expansion paths, and GPF consumer demands x D x.p; y/ imply linear Engel
curves (and income elasticilties tend to unity as total expenditure incuaser). These
assumptions may or may not be realistic over large changes in output or expenditure
(see Deaton and Muellbauer 1980, pp. 144–145, 151–153).
On the other hand it should be noted that Gorman Polar Form cost and indi-
rect utility functions are flexible functional forms. These former provide second
order approximations to arbitrary cost and indirect utility functions (Diewert 1980,
pp. 595–601). Therefore Gorman Polar Forms provide a local first order approx-
imation to any system of demand equations (except, of course, for cases such as
(6.38)).

6.3 Linear aggregation over agents using restrictions on the

distribution of output or expenditure

The analysis in the previous section was aimed at achieving consistent aggregation
over agents while imposing essentially zero restrictions on the distribution of output
or expenditure over agents. However these distribution are in fact highly restricted
in most cases, and these restrictions may help to achieve consistent aggregation.
Unfortunately there are few results on the relation between restrictions on the
distribution of “exogenous” variables such as output or income and consistent ag-
gregation. This section simply summaries several example that illustrate the effects
of alternative restrictions on the distribution of such variables.
First, output or expenditure may be choice variables for the agent rather than
exogenous variables, and it is very important to incorporate this fact into the analysis
of possibilities for consistent aggregation. For example, suppose that producers are
competitive profit maximizers and face the same output price p. Then the first order
condition in the output market for profit maximization is

@c f .w; y f /
Dp f D 1;


; F (6.40)
@y f
f
where c f .w; y f / D minx f N f f f
P
iD1 wi xi s.t. y .x / D y . This implies that the
marginal cost is identical across firms at all observed combinations of output levels
 
.y 1 ;


; y F / for given prices w, p.
Now remember from our earlier discussion that identical marginal cost across
firms is the condition for consistent linear aggregation of cost functions across firms
6.3 Linear aggregation over agents using restrictions on the distribution of output or expenditure
73

(irrespective of the distribution of output across firms) (see pages 64–66). Thus,
the assumption of competitive profit maximization and identical price imply that

an aggregate cost function C.w; f y f / exists over all profit maximizing lev-
P
P f P f
els of output f y D f y .w; p/. Moreover this aggregate cost function
C.w; f y f /, where f y f is restricted to the equilibrium levels f y f .w; p/,
P P P

also inherits the properties of cost minimization.4

As a second example, supposed that cost functions of individual consumers are
of the form

E f .p; uf / D af .p/uf C b f .p/ f D 1;


; F: (6.41)

this is slightly more general than the Gorman Polr Form (6.23) in the sense that
have the function af .p/ can vary over consumers. Nevertheless preference are still
quasi-homothetic.

Solving y f D af .p/uf C b f .p/ (6.41) for the consumer’s indirect utility
function yields

yf b f .p/
V f .p; y f / D f D 1;


; F; (6.42)
af .p/
and applying Roy’s Theorem to this result (6.42) yields
h
@b f .p/ f @af .p/ f f

i. f 2
@p
a .p/ @p
y b .p/ a .p/
xif .p; y f / D
i i
ı
1 af .p/
, (6.43)
@b f .p/ @af .p/  f f

D C y b .p/ af .p/
@pi @pi

Thus, in contract to the Gorman Polar Form,

xif .p; y f /
,
@af .p/
D af .p/ i D 1;


; N f D 1;


; F (6.44)
@y f @pi

i.e. Engel curves can vary over consumers (although these curves are still linear).
Now suppose that the distribution of “uncommitted expenditure” remains pro-
portionally constant over consumers, i.e. consumer incomes y 1 ;


; y F satisfy the
following restrictions for all variations in commodity prices p:

4
On the other hand, endogenizing consumer expedition y f , as the wage rate w times the amount
of labor supplied by the agent, is not sufficient for consistent linear aggregation in the case of
f
h y in this manneriimplies the additional first order condition
utility maximization. Endogenizing
  
@uf .x f ; x Le /=@x Le D @V f .p; y f /=@y f w where x Le  le of leisure for con-
 
sumer f , .f D 1;


; F /. Since the marginal utility of leisure @uf .x f ; x Le /=@x Le will

generally vary over consumers, the marginal utility of income @V f .p; y f /=@y f also varies
over consumers.
74 6 Aggregation Across Agents in Static Models
X 
F 
yf b f .p/ D f yf b f .p/ > 0 f D 1;


; F
f D1
XF (6.45)
f f
where  > 0,  D 1.
f D1

If the individual cost functions are of the form (6.41) and if the distribution of ex-
penditure over agents satisfies the restriction (6.45), then aggregate Marshallian de-
mand functions exist, have Gorman Polar Form and inherit the properties of utility
maximization.
Proof. Summing (6.43) over consumers and substituting in (6.45),
,
X @b f .p/ X @af .p/ hX  i
xf
X
D C f yf f
b .p/ af .p/
f i f @pi f @pi f

X @b f .p/ X @af .p/=@pi X f X @af .p/=@pi X f

D C f y f b .p/
f @pi f af .p/ f f af .p/ f
 
@ˇ.p/ @˛.p/=@pi X f 
D C y ˇ.p/ i D 1;


; N
@pi ˛.p/ f
(6.46)
f
where ˇ.p/  f b f .p/, ˛.p/  fFD1 af .p/ . Thus there exist aggregate
P Q
Marshallian demand functions of Gorman Polar Form. A GPF implies that aggregate
demands areP identical to those that could be generated by a single consumer with
expenditure f y f , and by assumption this consumer maximizes utility. Thus the
aggregate demand relations (6.46) inherit the properties of utility maximization. u
t
We have the following corollary to the above result.

Corollary 6.1. Suppose that the utility function of individual consumers are homo-
thetic, so that the expenditure function of individual consumers have the form

E f .p; uf / D af .p/uf f D 1;


; F
(6.47)

and also suppose that each consumer’s share f in total expenditure f y f is

P

fixed over the data set. Then aggregate Marshallian demand relations f x f D
P
P f
X.p; f y / exist, have the form
 X
@˛.p/=@pi
xif D
X
yf i D 1;


; N (6.48)
f ˛.p/ f

(using the notation of (6.46)) and inherit the properties of utility maximization.
In order to see that this is a special case of the result proves above, simply note
that (6.47) is a special case of (6.41) where b f .p/  0, and that b f .p/  0
.f D 1;


; F / reduce (6.45) to
6.4 Condition for exact nonlinear aggregation over agents 75

F
X
y f D f yf > 0 f D 1;


; F
f D1
(6.49)
F
X
where f > 0, f D 1
f D1

In one respect the assumption of homotheticity in (6.47) is more restrictive than

the quasi-homothetic Gorman Polar form (6.23) at constant price p the ratio of cost
minimizing demands is independent of the level of expenditure, i.e. increases in
expenditure y f lead to increases in consumption of all commodities along a ray
from the origin rather than from an arbitrary point. On the other hand, (6.47) is
more general than Gorman Polar Form (6.23) in the sense that the function af .p/
can vary over agent, so that (linear) Engel curves can vary over agents.

6.4 Condition for exact nonlinear aggregation over agents

The previous sections assumed that aggregate output or expenditure Y is to be con-

structed as a simple linear sum Y D f y f of the outputs of expenditure y f of
P
individual agents. More generally, the aggregation relation can be written as

Y D Y .y 1 ;


; y F /: (6.50)

Then the conditions for existence of an aggregate cost function for producers (for
example) can be written as
  X
C w; Y .y 1 ;


; y F / D c f .w; y f / (6.51)
f

for all .w; y 1 ;


; y F /. Differentiating (6.51) w.r.t. y f ,

@C.w; Y / @Y @c f .w; y f /
D f D 1;


; F (6.52)
@Y @y f @y f
which implies
,
@c f .w; y f / @c g .w; y g /

@Y @y
D f; g D 1;


; F: (6.53)
@y f @y g @y f @y g

Condition (6.53) implies that Y D Y .y 1 ;


; y F / is strongly separable (see Deaton

and Muellbauer 1980, pp. 137–142), so that
X 
Y DY hf .y f / : (6.54)
f
76 6 Aggregation Across Agents in Static Models

The corresponding aggregate cost function and aggregate demands are

  X 
C w; Y .y 1 ;


; y f / D ˛.w/ Y hf .y f / C ˇ.w/
f

1 F
 @˛.w/ X f f  @ˇ.w/
Xi w; Y .y ;


; y / D Y h .y / C i D 1;


; N:
@wi f @w i
(6.55)

The above cost function is more general than the Gorman Polar Form since the ag-
gregate output Y in not restricted to the linear case f y f . The aggregate cost func-
P
tion inherits the propertied of cost minimization, so the aggregate factor demands
X w; Y .y 1 ;


; y F / can be derived from the aggregate cost function using Shep-

ard’s Lemma. these conditions for exact nonlinear aggregation are less restrictive
than Gorman Polar Form, but it is not easy make this distinction operational (see
Deaton and Muellbauer 1980, pp. 154–158, for one attempt).

References

1. Deaton A. & Muellbauer, J. (1980). Economics and Consumer Behavior: pp. 144–145, 151–
153
2. Debreu, G. (1974). Excess demand functions, Journal of Mathematical Economics 1: pp. 15–
22
3. Diewert, W. E. (1977). Generalized slutsky conditions for aggregate consumer demand func-
tions, Journal of Economic Theory, Elsevier, vol. 15(2), pages 353-362, August.
4. Diewert, W. E. (1980). Symmetry Conditions for Market Demand Functions, Review of Eco-
nomic Studies: pp. 595–601
5. Mantel, R.(1977). Implications of Microeconomic Theory for Community Excess Demand
Function, pages 111–126 in M. D. Intriligator, ed., Frontiers of Quantitative Economics,
Vol. III A, North-Hollard
6. Sonnenschein, H. (1973). Do Walras’ identity and continuity characterize the class of com-
munity excess demand functions?. Journal of Economic Theory 6: pp. 345–354
Chapter 7
Aggregation Across Commodities: Non-index
Number Approaches

Consumers and also producers generally use a wide variety of commodities, so sub-
stantial aggregation (grouping) of commodities is necessary to make econometrics
studies manageable. This is particularly the case with flexible functional forms,
where the number of parameters to be estimated increases exponentially with the
number of commodities that are modeled explicitly. Presumably aggregation over
commodities generally misrepresent the choices faced by consumers and produc-
ers, so the microeconomic theory that applies to the true behavioral model with
disaggregated commodities may not generalize to a model with highly aggregated
commodities.
This leads to the following questions: when does aggregation over commodi-
ties not misrepresent the agent’s choices or behavior, and how is this aggregation
procedure defined? This lecture summaries two types of results related to this ques-
tion: the composite commodity theorem and conditions for two stage budgeting. The
composite commodity theorem of Hicks demonstrates that certain restrictions on the
covariation of prices permit consistent aggregation, and the discussion of two stage
budgeting shows that separability restrictions (plus other restrictions) on the struc-
ture of utility functions, production functions etc. also permit consistent aggregation
over commodities.1
The next lecture provides a more satisfactory answer to the above question. Cer-
tain index number formulas for aggregation over commodities can be rationalized
in terms of certain functional forms for production functions, cost functions, etc.,
including popular second order flexible functional forms. Thus certain index num-
ber formulas for aggregation over commodities inherit the desirable properties of
approximation that characterize the corresponding second order flexible functional
forms.

1
For simplicity we will assume a single agent, i.e. we will abstract from problems in aggregating
over agents.

77
78 7 Aggregation Across Commodities: Non-index Number Approaches

7.1 Composite commodity theorem

If the prices of several commodities are in fixed propositions over a data set, then
these commodities can be correctly treated as a single composite commodity with
one price. For example, suppose that a consumer maximizes utility over three com-
modities x1 ; x2 ; x3 and that prices p2 and p3 remain in fixed proportion over the
data set, i.e.

p2;t D  t p2;0 p3;t D  t p3;0 for all times t (7.1)

where p2;0 and p3;0 are base period prices of commodities 2 and 3, and  t is a
(positive) scalar that varies over time t. Equivalently the consumer can be viewed as
solving a cost minimization problem
3
X
E .p1 ; p2 ; p3 ; u/ D min pi xi
x (7.2)
i D1

s.t. u.x/ D u

Combining (7.1) and (7.2),

E .p1;t ; p2;t ; p3;t ; u t / D E .p1;t ;  t p2;0 ;  t p3;0 ; u t /

(7.3)
D EQ .p1;t ;  t ; u t / for all t

Cost minimizing behavior (7.2) implies that EQ .p1 ; ; u/ inherits all the properties
of a cost function, including Shephard’s Lemma:

Q 1 ; ; u/
@E.p
D x1
@p1
Q 1 ; ; u/
@E.p @ (7.4)
D .p1 x1 C  p2;0 x2 C  p3;0 x3 /
@ @
D p2;0 x2 C p3;0 x3

by the envelope theorem.

Thus p2;0 x2;t C p3;0 x3;t can be interpreted as the quantity x c of a composite
commodity at time t, and  t is the price of the corresponding composite commodity
at time t. The resulting system of Hicksian demands

x1 D x1h .p1 ; ; u/
(7.5)
x c D p2;0 x2 C p3;0 x3 D x c h .p1 ; ; u/

inherit the properties of cost minimizing demands. Therefore the corresponding sys-
tem of demands
x1 D x1 .p1 ; ; y/
(7.6)
x D p2;0 x2 C p3;0 x3 D x c .p1 ; ; y/
c
7.2 Homothetic weak separability and two-stage budgeting 79

inherit the properties of utility maximizing demands.

However, this composite commodity theorem may be of limited value in justify-
ing and defining aggregation over commodities. Even if two commodities are per-
ceived as relatively close substitutes in consumption, their relative prices may vary
significantly due to differences in the supply schedules of the two commodities.

7.2 Homothetic weak separability and two-stage budgeting

The assumption of “two-stage budgeting” has often been used to simplify studies of
consumer behavior. The general utility maximization problem (3.1) can be written
as

max u.x1A ;


; xN
A
A
; x1B ;


; xN
B
B
;


; x1Z ;


; xN
Z
Z
/  V .p; y/
x
NA
X NB
X NZ
X (7.7)
s.t. piA xiA C piB xiB C


C piZ xiZ D y
i D1 i D1 i D1

where x D .x1A ;


; xN
A
A
; x1B ;


; xN
B
B
;


; x1Z ;


; xN
Z
Z
/, i.e. there are NA C
NB C


C NZ commodities. Two-stage budgeting can then be outlined as follows.
In the first stage total expenditures y are allocated among broad groups of com-
modities x. For example the consumer decides to allocate the expenditures y A to
commodities x A D .x1A ;


; xNA
A
/, y B to commodities x B D .x1B ;


; xN B
B
/,


,
Z Z Z Z A B Z
y to commodities x D .x1 ;


; xNZ /, where y C y C


C y D y. This
allocation of expenditures among broad groups requires knowledge of total expen-
ditures y and of an aggregate price pQ A ; pQ B ;


; pQ Z for each group of commodities.
Thus in the first stage a consumer is viewed as solving a problem of the form

max u.xQ A ; xQ B ;


; xQ Z /
xQ (7.8)
s.t. pQ A xQ A C pQ B xQ B C


C pQ Z xQ Z D y

where xQ D .xQ A ;


; xQ Z / and pQ D .pQ A ;


; pQ Z / denote aggregate commodities

and prices for the broad groups A;


; Z. Alternatively, utilizing the equivalence
between utility maximization and cost minimization (see pages 27 to 28), in the first
stage a consumer can be viewed as solving a cost minimization problem of the form

min pQ A xQ A C pQ B xQ B C


C pQ Z xQ Z
xQ (7.9)
s.t. Q D u
u.x/

In the second stage the group expenditure y A ; y B ;


; y Z are allocated among

the commodities within each group. Here the consumer solves maximization prob-
lems of the form
80 7 Aggregation Across Commodities: Non-index Number Approaches

max uA .x1A ;


; xN
A
A
/


max uZ .x1Z ;


; xN
Z
Z
/
xA xZ
NA
X NZ
X (7.10)
s.t. piA xiA D y A


s.t. piZ xiZ D y Z
i D1 i D1

A
where uA .x1A ;


; xNA
Z
/;


; uZ .x1Z ;


; xNZ
/ are interpreted as “sub-utility func-
tion” for the commodities within each group A;


; Z.
What restrictions on the consumer’s utility function u.x/ imply that two-stage
budgeting is realistic, i.e. under what restrictions do the general utility maximiza-
tion problem (7.7) and the two-stage procedure (7.8) and (7.10) yield the same so-
lutions x  ? First consider the second stage of two-stage budgeting. Define a weakly
separable utility function as follows:
Definition 7.1. A utility function u.x/ is defined as “weakly separable” in commod-
ity groups x A D .x1A ;


; xN
A
A
/;


; x Z D .x1Z ;


; xN Z
Z
/ if and only if u.x/ can
be written as
h i
u.x/ D uQ uA .x1A ;


; xN A
A
/;


; uZ
.x Z
1 ;


; xN
Z
Z
/

Q A ;


; uZ / and sub-utility func-

over all x for some macro-utility function u D u.u
A A A A Z Z Z Z
tion u D u .x1 ;


; xNA /;


; u D u .x1 ;


; xNZ
/.
The second stage of two-stage budgeting is equivalent to the restriction that the
consumer’s utility function u.x/ is weakly separable. To be more precise,

Property 7.1. The general utility maximization problem (7.7) and a series of
“second stage” maximization problems (7.10) (conditional on group expen-
diture y A ;


; y Z ) yield the same solution x  if and only if u.x/ is weakly
separable in the above manner in Definition 7.1.

Proof. First, suppose that u.x/ is weakly separable as in Definition 7.1 and that
A Z
Q
@u=@u > 0;


; @u=@u
Q > 0. Then utility maximization (7.7) requires that
A Z
each subutility u ;


; u be maximized conditional on its group expenditure
y A ;


; y Z . For example if x A solving Definition 7.1 does not solve uA .x A /
s.t. N
P A A A A A A
iD1 pi xi D y , then y could be reallocated among commodities x so
A B Z
as to increase u without decreasing u ;


; u , i.e. so as to increase the total util-
ity level u without violating the budget constraint px D y. Thus the second stage
of two-stage budgeting is satisfied if u.x/ is weakly separable. Second, suppose
that two-stage budgeting is satisfied. two-stage budgeting implies x A D x A .p; y/
solving (7.7) can be written as

xiA D xiA .p A ; y A / i D 1;


; NA (7.11)

and similarly for subgroups B;


; Z. Without loss of generality x A solves

7.2 Homothetic weak separability and two-stage budgeting 81
 
max u x A ; x B ;


; x Z
xA
NA
X (7.12)
s.t. piA xiA Dy A

iD1

where x B ;


; x Z are fixed at their equilibrium levels solving (7.7), (7.11) im-
plies that (given p A ; y A ) x A is independent of p B ;


; p Z and hence indepen-
dent of x B ;


; x Z . Thus (7.12) reduces to (7.10), i.e. there exists a subutil-
ity function uA .x A / for commodity group A that is independent of the levels of
other commodities x B ;


;
x Z . Thus two-stage budgeting implies weak separabil-
ity u uA .x A /;


; uZ .x Z / . ut

Second, consider the first stage of two-stage budgeting. The critical point here
is to be able to construct a price pQ A ;


; pQ Z for each commodity group A;


; Z
such that a first stage utility maximization or cost minimization problem yields the
optimal allocation of expenditure across subgroups A;


; Z.
Given weak separability of u.x/, a sufficient condition for the first stage is ho-
motheticity of each subutility function uA .x A /;


; uZ .x Z /. To be more precise,

Property 7.2. If u.x/ is weakly separable and each subutility function

uA .x A /;


; uZ .x Z / is homothetic, then there exists a first stage maximization
problem that obtains the same distribution of group expenditures y A ;


; y Z
as in the general case (7.7).

Proof. Weak separability of u.x/ implies second stage maximization (7.10) and
equivalently a series of cost minimization problems.
NA
X
min piA xiA D E A .p A ; uA /
xA
i D1

s.t. uA .x A / D uA
:: (7.13)
:
NZ
X
min piZ xiZ D E Z .p Z ; uZ /
xZ
i D1

s.t. uZ .x Z / D uZ

where uA D uA .x A /;


; uZ D uZ .x Z / for x A ;


; x Z solving (7.7). Un-

der weak separability the consumer’s general cost minimization problem E.p; u / D
minx px s.t. u.x/ D u can be restated as
82 7 Aggregation Across Commodities: Non-index Number Approaches
   
E p; u D E A p A ; uA C


C E Z p Z ; uZ


min
uA ;


;uZ
  (7.14)
s.t. uQ uA ;


; uZ D u

uA .x A / homothetic implies E A .p A ; uA / D .uA /E A .p A / (see Property 1.3.a on

page 6) where without loss of generality we can define .uA / D uA (constant
returns to scale), since the indexing of the indifference curves representing the con-
sumer’s preferences is arbitrary. Therefore under homotheticity (7.14) reduces to
   
E p; u D min uA e A p A C


C uZ e Z p Z


uA ;


;uZ
  (7.15)
s.t. uQ uA ;


; uZ D u

This can be interpreted as a first stage cost minimization problem with aggregate
prices c A .p A /;


; c Z .p Z / and leading to the following optimal allocation of ex-
penditures y over subgroups:

y A D uA e A .p A /;


; y Z D uZ e Z .p Z /

Weak separability places substantial restrictions on the degree of substitution be-

tween commodities in different groups. Since weak separability implies that utility
maximizing demands for a group of commodities x A depends only on prices p A
and group expenditure y A (7.11), it follows that other prices p B influence demands
x A only through changes in the optimal level of expenditure y A for the group. This
implies the following restriction on the Hicksian substitution effects across groups:

@xiA .p; u / @xiA .p A ; y A / @xjB .p B ; y B /

D ' AB .p; y/
@pjB @y A @y B (7.16)
i D 1;


; NA j D 1;


; NB

where the function ' AB .p; y/ is independent of the choice of commodities i; j

from group A; B (see Deaton and Muellbauer, pp. 128–129 for a sketch of a proof),
moreover, this relationship (7.16) is both necessary and sufficient for weak separa-
bility of groups A and B, i.e. restrictions (7.16) exhaust the implications of weak
separability.
The most obvious application of this discussion of two-stage budgeting is in the
estimation of, e.g., consumer demand for food. In general the demand for food goods
x A D .x1A ;


; xN
A
A
/ will depend on the prices of all food and non-food final goods
and on total expenditure y. On the other hand suppose that the consumer’s utility
function u.x/ is weakly separable in food commodities x A and all other commodi-
B A A B B


ties x , i.e. u.x/ D uQ u .x /; u .x / . Then the demand equations for food
xiA D xiA p A ; p B ; y .i D 1;


; NA / can be simplified to xiA D xiA p A ; y A

.i D 1;


; NA / (7.11) which can be derived from a subutility maximization prob-

lem
7.3 Implicit separability and two-stage budgeting 83

max uA .x A / D V A .p A ; y A /
xA
NA
X (7.17)
s.t. piA xiA D y A
i D1
A A A
Here V .p ; y / has the properties of an indirect utility function and the corre-
sponding Marshallian demands x A D x A .p A ; y A / inherit the properties of utility
maximization.
One complication in estimating the above model (7.17) for food demand is that
food expenditure y A is not exogenous to the consumer, and ignoring this fact gen-
erally leads to biases in estimation. In the absence of any further assumptions (be-
yond weak separability) y A generally depends on all prices and total expenditure,
i.e. y A D y A .p A ; p B ; y/.
However if the subutility functions uA .x A / and uB .x B / are homothetic (im-
plying two-stage budgeting), then the corresponding expenditure functions can be
written as uA e A .p A / and uB e B .p B / and the relation y A D y A .p A ; p B ; y/ can be
rewritten as h i
y A D y A e A .p A /; e B .p B /; y (7.18)

Thus we can posit homothetic functional forms E A D uA e A .p A / and E B D

uB e B .p B /, derive the corresponding indirect utility function V A D y A =e A .p A /
and differentiate this indirect utility function to obtain the estimating equations
x A D x A .p A ; y A /, and use these functional forms e A .p A /; e B .p B / as the aggre-
gate prices in (7.18). Unfortunately the expenditure equation (7.18) still depends on
prices p B as well as p A , but at least the estimating equation (7.18) is more restricted
than the general equation y A D y A .p A ; p B ; y/.2

7.3 Implicit separability and two-stage budgeting

An alternative two-stage budgeting procedure can be obtained if certain separability

restrictions hold for the consumer’s cost function E.p; u/ rather than for his utility
function u.x/. Preferences are defined as “implicitly separable” in broad groups
A;


; Z if the cost function can be written in the form
h i
E.p; u/ D EQ e A .p A ; u/;


; e Z .p Z ; u/; u (7.19)

where p A D .p1A ;


; pN A
A
/;


; p Z D .p1Z ;


; pN
Z
Z
/. Note that total util-
A Z
ity u (rather than subutilities u ;


; u ) appear in each of the cost function
e A .p A ; u/;


; e Z .p Z ; u/, so there are no group subutilities in contrast to the case
where u.x/ is weakly separable.

2
Alternatively we can eliminate y A from the demand equations x A D x A .p A ; y A / using the
PNA A A
identity y A  i D1 pi x i .
84 7 Aggregation Across Commodities: Non-index Number Approaches

The following two-stage budgeting procedure is defined by simple differentiation

Q A ;


; e Z / and then the group cost function e A .p A ; u/;


; e Z .p Z ; u/:
of the macrofunction E.e

PNA
piA xiA Q A ;


; e Z ; u/
@ log E.e
A iD1
(first stage) s  D
y @ log e A
:: (7.20a)
:
PNZ
piZ xiZ Q A ;


; e Z ; u/
@ log E.e
sZ  iD1
D
y @ log e Z

piA xiA @ log e A .p A ; u/

(second stage) siA  D i D 1;


; NA
yA @ log piA
:: (7.20b)
:
piZ xiZ @ log e Z .p Z ; u/
siZ  D i D 1;


; NZ
yZ @ log piZ

where y  y A C


C y Z (i.e. total expenditure) and y A  N

P A A A
iD1 pi xi (expendi-
ture of group A), etc. Thus the budget shares s A ;


; s Z of groups A;


; Z are ob-
tained by simple logarithmic differentiation of the macrofunction E.e Q A ;


; e Z ; u/,
A Z
and the share si .i D 1;


; NA /,


, si .i D 1;


; NZ / of each commodity in
group expenditure is obtained by simple logarithmic differentiation of the group
cost functions e A .p A ; u/;


; e Z .p Z ; u/.

Proof. Differentiating (7.19) w.r.t. piA and applying Shephard’s Lemma,

@E.p; u/ Q @e A.
/
@E.
/
D
@piA @e A @piA (7.21)
D xiA .p; u/ i D 1;


; NA ;

Thus
NA
X
A
y  xiA piA
iD1
N
Q X
@E.
/ A
@e A .
/ A (7.22)
D pi
@e A @piA iD1
Q
@E.
/
D e A .
/ by Euler’s Theorem
@e A
References 85

since e A .p A ; u/ is linear homogeneous in p A .3

yA
sA 
y
Q
@E.
/=@e A
D (7.23)
y=e A .
/
Q
@ log E.
/
D
@ log e A

@EQ .
/ yA @e A .
/ @EQ .
/

which is (7.20a). Substituting @e A
D e A .
/
(7.22) into @piA @e A
D xiA (7.21)
yields (7.20b).

Note that implicit separability is sufficient for the two-stage budgeting procedure
outlined in (7.20). In contrast weak separability of u.x/ was sufficient only for the
second stage of the budgeting procedure discussed in the previous section. Also
note that implicit separability implies that the ratio of commodities within a group
is independent of prices of commodities outside the group, i.e.
h i
@ xiA .p; u/=xjA .p; u/
D0 i; j D 1;


; NA k D 1;


; NB (7.24)
@pkB

(see (7.21)). In contrast, weak separability of u.x/ implied that the ratio of marginal
rates of substitution between commodities within a group is independent of levels
of commodities outside the group.

References

1. Deaton A. & Muellbauer, J. (1980). Economics and Consumer Behavior

3
E.p; u/ is linear homogeneous in p only if e A .p A ; u/;


; e Z .p Z ; u/ are also linear
homogeneous in prices and EQ .e A ;


; e Z ; u/ is linear homogeneous in e A ;


; e Z . How-
ever note that e A .p A ; u/ does not equal total expenditure y A on group A (see (7.22)). Thus
c A .p A ; u/;


; c Z .p Z ; u/ are to be interpreted as group price indexes that depend on utility
level u.
Chapter 8
Index Numbers and Flexible Functional Forms

In this lecture we show that particular index number formulas for aggregating over
commodities can be rationalized in terms of particular functional forms for produc-
tion functions or dual cost or profit functions. Approximately correct procedures for
aggregating over commodities are presented for cases of Translog and Generalized
Leontief functional forms. Since these functional forms provide a second order ap-
proximation to any true form, the corresponding index number formulas can often
be interpreted as approximately correct.
The results obtained here should be contrasted with the previous lecture. There
consistent aggregation over commodities was rationalized essentially in terms of
assumptions of separability between groups of commodities. Here specific aggre-
gation procedures are rationalized essentially in terms of specific functional forms
for production functions or dual cost functions. Since assumptions of Translog or
Generalized Leontief functional forms are usually considered less restrictive than
assumptions of (homothetic) weak separability, this lecture presents a more useful
basis for a theory of approximate aggregation over commodities.

8.1 Laspeyres index numbers and linear functional forms

Until recently most index number computations have used simple base period
weighting schemes, and the most common of these are Laspeyres quantity and price
indexes. The Laspeyres quantity index can be written as
PN
X1 pi;0 xi;1
D PiND1 (8.1)
X0 i D1 pi;0 xi;0

where p0 D .p1;0 ;


; pN;0 / denotes the prices for the N commodities in the base
period .t D 0/, x0 D .x1;0 ;


; xN;0 / denotes the quantities of the N commodities
in the base period .t D 0/, and x1 D.x1;1 ;


; xN;1 / denotes the quantities of the
N commodities in any other period t D 1.

87
88 8 Index Numbers and Flexible Functional Forms

This aggregation procedure (8.1) can be defined as correct if the ratio of aggre-
gates X1 , X0 provides an accurate measure of the contributions of inputs 1;


; N
to the producer’s output in different time periods. Thus in the case where aggrega-
tion is defined over all inputs (1;


; N is to be interpreted as all inputs) and there is
a single output, the quantity index X1 =X0 in (8.1) is correct if X1 =X0 is equal to the
ratio of outputs f .x1;1 ;


; xN;1 /=f .x1;0 ;


; xN;0 / for any time periods t D 0; 1.
Similarly a Laspeyres price index can be written as
PN
P1 xi;0 pi;1
D PiD1
N
(8.2)
P0 iD1 xi;0 pi;0

where the prices p D .p1 ;


; pN / for any period are weighted by the base period
quantities x0 D .x1;0 ;


; xN;0 /. Interpreting commodities 1;


; N as inputs in
production and assuming cost minimizing behavior, this aggregation procedure can
be defined as correct if the ratio of the aggregates P1 , P0 provides an accurate
measure of the contribution of inputs 1;


; N to the cost of attaining a given level
of output y. In the case where aggregation is defined over all inputs (1;


; N is to
be interpreted as all inputs), the price index P1 =P0 is correct if P1 =P0 is equal to
the ratio of minimum costs C.p1;1 ;


; pN;1 ; y/=C.p1;0 ;


; pN;0 ; y/ for any two
time periods and a common output level y.
Are the above aggregation procedures (8.1)–(8.2) correct for some cases of pro-
duction functions? The answer is yes: assuming static competitive profit maximizing
or cost minimizing behavior for a firm, an aggregation procedure (8.1) or (8.2) over
inputs x D .x1 ;


; xN / can be rationalized in terms of a linear production function
with a family of parallel straight line isoquants

x2

0 x1 (8.3)

and also in terms of a linear production function with fixed coefficients, i.e. right
angle isoquants
8.2 Exact indexes for Translog functional forms 89

x2

0 x1 (8.4)
The first case (8.3) assumes perfect substitution between inputs and the second case
(8.4) assumes zero substitution between inputs.
More formally, given Laspeyres quantity and price indexes X1 =X0 (8.1), a linear
production function y D f .x/ satisfying either (8.3) or (8.4), and static competitive
profit maximizing or cost minimizing behavior, then

X1 f .x1;1 ;


; xN;1 /
D
X0 f .x1;0 ;


; xN;0 /
(8.5)
P1 c.p1;1 ;


; pN;1 /
D for all t D 1; 0
P0 c.p1;0 ;


; pN;0 /

where c.p/ denotes a unite cost function (the minimum cost of producing one unite
of output y given prices p) (see Diewert 1976 pp. 182–183 for a proof of (8.5) ).
Under the above assumptions X1 =X0 is the ratio of output and P1 =P0 is the ratio
of unite cost of output for the two periods t D 1; 0. In this sense Laspeyres quantity
and price indexes are exact for both a linear production function satisfying (8.3) and
a linear production function satisfying (8.4).
In either case linear production functions (with either zero or perfect substitu-
tion between inputs are very unrealistic and highly restrictive. Linear production
functions can provide only a first order approximation to an arbitrary production
function.1 This suggests that Laspeyres indexes may often lead to substantial er-
rors in aggregation over commodities, and it is unlikely that aggregate data col-
lected in this manner inherits properties of profits maximization or cost minimiza-
tion from disaggregate data. In the next two sections we obtain more positive results
by demonstrating that particular quantity and price indexes are correct for Translog
and Generalized Leontief functional forms.

8.2 Exact indexes for Translog functional forms

Economists have frequently advocated the use of Divisia indexes rather than Laspeyres
indexes for aggregating over commodities (e.g. Hulten 1973, pp. 1017–1026). A

1
It can easily be shown that Laspeyres indexes also provide a first order approximation to an
arbitrary (true) index (see Deaton1980, pp.173–174).
90 8 Index Numbers and Flexible Functional Forms

Divisia quantity index

Z for
X(e.g.) inputs can be defined in continues time by the line
Xt N @xi .t /=@t
integral D exp si .t / where si .t /  PNwi .t /xi .t / :
X0 iD1 xi .t / j D1 wj .t /xj .t /
The following Törnqvist index is the most commonly used discrete approximation
to the Divisia quantity index for inputs:
  N  
X1 X xi;1
log D si log
X0 xi;0
iD1
! (8.6)
1 wi;1 xi;1 wi;0 xi;0
where si  PN C PN i D 1;


; N:
2 j D1 wj;1 xj;1 j D1 wj;0 xj;0

The Törnqvist quantity index (8.6) is exact for a Translog production function
with constant returns to scale. In other words, assuming static competitive cost min-
imizing behavior,    
Xt f .x t /
log D log (8.7)
Xs f .xs /
for all periods s, t when f .x/ is Translog constant returns to scale and the quantity
index is calculated as in (8.6). Such an index, which is exact for a constant returns
(or variable returns) to scale flexible form for f .x/, is termed superlative.

Proof. First we establish an algebraic result for a quadratic function

X 1X X
g.´/ D a0 C ai ´ i C aij ´i ´j .aij D aj i for all i; j /
i 2 i j

or, in matrix notation,

1
g.z/ D a0 C aT z C zT A z .A symmetric/ (8.8)
2
Then
1 1 T
g.z1 / g.z0 / D aT .z1 z0 / C zT1 A z1 z A z0
2 2 0
1 1
D aT .z1 z0 / C zT1 A.z1 z0 / C zT0 A.z1 z0 /
2 2 (8.9)
since A is symmetric
1
D .a C Az1 C a C Az0 /T .z1 z0 /
2
which implies

@g.z0 / T
 
1 @g.z1 /
g.z1 / g.z0 / D C .z1 z0 / (8.10)
2 @z @z
8.2 Exact indexes for Translog functional forms 91

Moreover (8.10) implies (8.8), i.e. (8.8) is correct if and only if (8.10) is
correct. Since a Translog production function log f D a0 C N
P
i D1 ai log xi C
PN PN @ log f .x/ @f .x/=@xi
iD1 j D1 aij log xi log xj is quadratic in logs, (8.10) and @ log xi D f .x/=xi
.i D 1;


; N / imply
N    
1 X @f .x1 / xi;1 @f .x0 / xi;0 xi;1
log f .x1 / log f .x0 / D C log :
2 @xi;1 f .x1 / @xi;0 f .x0 / xi;0
iD1
(8.11)
wi
Static competitive cost minimization implies @f@x.x/
i
D @C.w;y/=@y
.i D 1;


; N /,
PN @f .x/
and constant returns to scale implies f .x/ D i D1 @xi
xi (Euler’s theorem).
Substituting these to (8.11),
N
  !  
f .x1 / 1X wi;1 xi;1 wi;0 xi;0 xi;1
log D PN C PN log (8.12)
f .x0 / 2 j D1 wj;1 xj;1 j D1 wj;0 xj;0
xi;0
i D1

where right hand side is the Törnqvist input quantity index (8.6).

Moreover, the Törnqvist quantity index (8.6) is exact only for a Translog con-
stant returns to scale production function (this follows from the equivalence between
(8.10) and a quadratic function g.z/ (8.8)).
The assumption of a constant returns to scale production function appears to be
crucial to the interpretation of the particular Törnqvist quantity index (8.6) as exact.
This assumption is not crucial only in the case of pair-wise comparisons of aggregate
inputs, i.e. if the aggregate index (8.6) is to be calculated only for two time periods
t D 0; 1 (see Diewert 1976, Op. cit.).
Nevertheless, a quantity index closely related to (8.6) can be interpreted as exact
for a general Translog production function. Define the following quantity index:
  N    
X1 1 X wi;1 xi;1 wi;0 xi;0 xi;1
log D C log (8.13)
X0 2 py;1 f .x1 / py;0 f .x0 / xi;0
iD1

where py;t  price of the (single) output in period t . This deviates from the
Törnqvist quantity index (8.6) only in that wi;t xi;t is divided by total revenue
py;t f .x t / rather than by total cost N
P
i D1 wi;t xi;t . In the case of constant returns to
scale f .x/ D N
P @f .x/ PN
iD1 @xi
x i (Euler’s theorem) and in turn py f .x/ D i D1 wi xi .
Thus (8.11) reduces to the Törnqvist index (8.6) in the case of constant returns to
scale and competitive profit maximization. Unlike the index (8.6), the above quan-
tity index (8.13) is exact for a general (variable returns to scale) Translog production
function. This result requires the assumption of profit maximization rather than sim-
ple cost minimization, in contrast to (8.6).

Proof. Substituting the first order conditions for static competitive profit maximiza-
wi;t
tion @f@x.xit / D py;t .i D 1;


; N / into (8.11),
92 8 Index Numbers and Flexible Functional Forms
  N    
f .x1 / 1 X wi;1 xi;1 wi;0 xi;0 xi;1
log D C log (8.14)
f .x0 / 2 py;1 f .x1 / py;0 f .x0 / xi;0
i D1

i.e. (8.13) is exact for the general Translog f .x/.

This index (8.13) can be interpreted as an alternative Törnqvist index approximating

the continuous Divisia quantity index.
Given an input quantity index X1 =X0 such as (8.6) or (8.13), we can easily con-
struct a corresponding implicit input price index using the following factor reversal

B
equation:
   PN
W1 X1 wi;1 xi;1
D PiD1N
(8.15)
W0 X0 iD1 wi;0 xi;0
i.e. the product of the input price and quantity indexes is equal to the ratio of total
expenditure on the N disaggregate inputs for the corresponding time periods t D
0; 1. Assuming X1 =X0 D f .x1 /=f .x0 / (8.7), (8.15) implies

B

W

D
1
P N
iD1 wi;1 xi;1 =f .x1 /
(8.16)
PN
W0 iD1 wi;0 xi;0 =f .x0 /

C
i.e. .W1 =W0 / can be interpreted as the index of average costs of production. Obvi-
ously if a quantity index is superlative then the corresponding implicit price index
is superlative in an analogous manner.
Alternatively an input price index can be calculated directly rather than by using
(8.15). Now assume that the production function is constant returns to scale and that
the cost function C.w; y/ D y c.w/ is Translog: log c.w/ D a0 C N
P
PN PN iD1 ai log wi C
iD1 a
j D1 ij log w i log w j . Define the following Törnqvist price index for in-
puts:
  X N  
W1 wi;1
log D si log (8.17)
W0 wi;0
iD1

where s D .s1 ;


; sN / are calculated as in (8.6).2 Proceeding as in the proof of

(8.7), we obtain the result
   
W1 c.w1;1 ;


; wN;1 /
log D log (8.18)
W0 c.w1;0 ;


; wN;0 /

which is analogous to (8.16). Thus the Törnqvist price index (8.17) is exact for a
Translog unit cost function. In the absence of constant returns to scale in production,
the assumption of a Translog cost function C.w; y/ implies

2
Since flexible functional forms are not self-dual (e.g. a Translog production function does not
imply a Translog cost function, or vice-versa), the price index (8.17) for a Translog unit cost func-
tion is not equivalent to the implicit price index (8.16) for a Translog constant returns to scale
production function.
8.2 Exact indexes for Translog functional forms 93
  N  
W1 X wi;1
log D si log
W0 wi;0
i D1 (8.19)
 
C.w1;1 ;


; wN;1 ; y1 /
D log
C.w1;0 ;


; wN;0 ; y0 /

i.e. the Törnqvist index (8.17) is equal to the ratio of total costs in different time pe-
riods t D 0; 1: Thus, in the absence of constant returns to scale, the direct Törnqvist
index (8.17) cannot strictly be interpreted as a price index for inputs since it depends

C
upon a measure y1 ; y0 of input quantities x1 ; x0 as well as upon input prices w1 ; w0 .
In contrast, consider the index .W1 =W0 / calculated implicitly from (8.15) using
the quantity index (8.13), science this index satisfies (8.16) for a general Translog
production function and profit maximizing behavior, it can be interpreted as the ratio
of average costs of production

log
B

W

1
D log

C.w 1;1 ;



; wN;1 ; y1 /=y1

(8.20)
W0 C.w1;0 ;


; wN;0 ; y0 /=y0

even in the absence of constant returns to scale.

The derivations of exact index number equations are complicated somewhat in
the case of multiple outputs y D .y1 ;


; yM /. Generalizations of the Translog
production function do not appear to provide an entirely satisfactory basis for index
number formulas. For example, assume a Translog transformation function
T
X T X
X T
log y1 D a0 C ai log ´i C aij log ´i log ´j (8.21)
iD1 iD1 j D1

where z  .y2 ;


; yM ; x1 ;


; xN / .T D M 1 C N / and static competitive

profit maximization
XM X N
max pi yi wi xi
y;x (8.22)
iD1 i D1
s.t. y1 D y1 .z/
Then the quadratic identity (8.11) implies
  N   M  
yi;1 xi;1 yi;1
siy log
X X
log D six log
yi;0 xi;0 yi;0
i D1 iD2
 
1 wi;1 xi;1 wi;0 xi;0 (8.23)
where six  C
2 pi;1 yi;1 pi;0 yi;0
 
1 p y
i;1 i;1 p i;0 yi;0
siy  C
2 pi;1 yi;1 pi;0 yi;0
94 8 Index Numbers and Flexible Functional Forms

(8.23) defines quantity indexes that are exact for the Translog transformation func-
y
tion (8.21). However the output quantity index M
P
i D2 si log .yi;1 =yi;0 / incorporates
the level of the first output y0 only indirectly via the definitions of the weights siy
for output 2;


; M .
A more satisfactory approach to the derivation of index numbers in the case of
multiple outputs may be in terms of the cost function C.w; y/. Assume a Translog
joint cost function
N
X M
X
log C D a0 C ai log wi C bi log yi
iD1 iD1
N X
X N M X
X M
C aij log wi log wj C bij log yi log yj (8.24)
iD1 i D1 i D1 iD1
T X
X T
C cij log wi log yj
iD1 i D1

and static competitive profit maximizing behavior. Then the quadratic identity
(8.11), Shephard’s Lemma and @C.w; y/=@yi D pi .i D 1;


; M / imply
  N   M 
C1 wi;1 yi;1
siy
X X
log D six log C log
C0 wi;0 yi;0
i D1 iD1
!
1 wi;1 xi;1 wi;0 xi;0
where six  PN C PN (8.25)
2 iD1 wi;1 xi;1 iD1 wi;0 xi;0
!
1 pi;1 yi;1 pi;0 yi;0
siy  PN C PN :
2 iD1 wi;1 xi;1 iD1 wi;0 xi;0

The first sum N x

P
i D1 si log .wi;1 =wi;0 / can be interpreted as a Törnqvist price
y
index log .W1 =W0 / for all inputs, and the second sum M
P
iD1 si log .yi;1 =yi;0 / can
be interpreted as a Törnqvist quantity index log .Y1 =Y0 / for all outputs. In order to
see this, rewrite equation (8.25) as
   
C1 W1 Y1
D

C0 W0 Y0
  X  
W1 N
x wi;1
where log D s log (8.26)
W0 i D1 i wi;0
  X  
Y1 N yi;1
log D s y log
Y0 i D1 i yi;0

i.e. the index of total costs is always equal to the product of the input price index and
the output quantity index. This confirms that Y1 =Y0 can be interpreted as an output
8.2 Exact indexes for Translog functional forms 95

quantity index and W1 =W0 can be interpreted as an index of the contributions of

inputs to the average cost of aggregate output.
The corresponding implicit Törnqvist input quantity and output price indexes can
then be calculated from the equations

W1
 

X1
D
B PN
iD1 wi;1 xi;1
PN
W0 X0 wi;0 xi;0
A
iD1
   PM (8.27)
P1 Y1 iD1 pi;1 yi;1
D PM
P0 Y0 iD1 pi;0 yi;0

B B
These implicit indexes .X1 =X0 / and .P1 =P0 / are also superlative.3 Finally, con-
sider the problem of deriving index numbers for aggregation of commodities in the
case of consumer behavior. Assuming that the utility function u.x/ is homothetic (or
equivalently constant returns to scale) and that the unit cost function e.p/ D min px
x
s.t. u.x/ D 1 is Translog, then the Törnqvist price index
N
  !  
P1 1X pi;1 xi;1 pi;0 xi;0 pi;1
log D PN C PN log (8.28)
P0 2 j D1 pj;1 xj;1 j D1 pj;0 xj;0
pi;0
i D1

and the corresponding implicit quantity index

log
B

X

1
D log
PN
i D1 pi;1 xi;1
!
log

P1

(8.29)
PN
X0 j D1 pj;0 xj;0 P0

are exact (see the discussion of (8.17)–(8.18)).

Alternatively suppose that the consumer’s utility function u.x/ is Translog as
well as homothetic:
N
X N X
X N
log u D a0 C ai log xi C aij log xi log xj (8.30)
i D1 i D1 j D1

u.x/ constant returns to scale implies that the indirect utility function V .p; y/ is
constant returns to scale in expenditure y, so that (using Euler’s theorem) V .p; y/ D
@V .p;y/
@y

y. Applying the quadratic identity (8.10) to (8.30), and using the first order
conditions @u.x/
@xi
D @V .p;y/
@y

pi .i D 1;


; N / and in turn @V .p; y/=@y D u=y,
we obtain the result

3
In the case of constant returns to scale in production, W1 =W0 can also be interpreted as a price
index of the contributions of inputs to the marginal cost of aggregate output.
96 8 Index Numbers and Flexible Functional Forms
  N  
u.x1 / X xi;1
log D si log
u.x0 / xi;0
i D1
! (8.31)
1 pi;1 xi;1 pi;0 xi;0
where si  PN C PN
2 j D1 pj;1 xj;1 j D1 pj;0 xj;0

The right hand side of (8.31) defines a Törnqvist quantity index log.X1 =X0 / which
is exact for a homothetic Translog utility function u.x/, and the corresponding im-
plicity price index can be calculated using (8.29).

8.3 Exact indexes for Generalized Leontief functional forms

Results in the previous section demonstrated that many index numbers closely
related to popular Törnqvist indexes can be rationalized in terms of underlying
Translog functional forms. This section briefly illustrates that different index num-
ber formulas are implied by Generalized Leontief functional forms. Nevertheless,
since both classes of functional forms provide a second order approximation to a
true form, the various index number formulas should lead to similar results at least
for small changes in quantities and prices.
First, assume a constant returns to scale generalized Leontief production function
N X
N
X p p
f .x/ D aij xi xj (8.32)
iD1 j D1

and static competitive profit maximization. This implies

PN 1=2
f .x1 / iD1 si;0 .xi;1 =xi;0 /
D PN 1=2
f .x0 / j D1 sj;1 .xj;0 =xj;1 / (8.33)
wi;t xi;t
where si;t  PN
kD1 wk;t xk;t

i.e. the quantity index number X1 =X0 corresponding to the right hand side of (8.33)
is exact for the production function (8.32).

Proof. Competitive profit maximization and constant returns to scale imply p @f@x.x/
i
D
PN
wi and pf .x/ D iD1 wi xi , so that
wi;t
vi;t  PN
j D1 wj;t xj;t
(8.34)
@f .x t /=@xi;t
D .i D 1;


; N /
f .x t /
8.4 Two-stage aggregation with superlative index numbers 97

for all time periods t. Substituting the derivatives of (8.32) into (8.34),
1=2 PN p
.xi;t / j D1 aij xj;t
vi;t D (8.35)
f .x t /
p p
multiplying vi;0 by xi;1 xi;0 and summing over i D 1;


; N ,

N PN PN p p
X p p iD1 j D1 xi;1
aij
xi;0
xi;1
vi;0
xi;0 D (8.36)
f .x0 /
i D1

and similarly,
N PN PN p p
X p p i D1 j D1 xi;0
aij
xj;1
xi;0
vi;1
xi;1 D (8.37)
f .x1 /
iD1

p p p
Dividing (8.36) by (8.37) (noting ai;j D aj;i and xi;1 xi;0 D xi;1 =xi;0
xi;0 ),

.xi;1 =xi;0 /1=2
vi;0
xi;0

P
i f .x1 /

1=2 D : (8.38)
P
xj;0 =xj;1
vj;1
xj;1 f .x0 /
j

Alternatively , assume constant returns to scale in production and a Generalized

Leontief unit cost function c.w/ D C.w; y/=y:
N
X p p
c.w/ D aij wi wj (8.39)
i D1

These assumptions and competitive profit maximization imply

PN
c.w1 / si;0 .wi;1 =wi;0 /1=2
D P i D1
1=2 (8.40)
c.w0 / N
sj;1 wj;0 =wj;1
j D1

where s0 , s1 are defined as in (8.33) (the proof is analogous to the above proof of
(8.33)). Thus the above price index for inputs is exact for a Generalized Leontief
unit cost function (8.39).

8.4 Two-stage aggregation with superlative index numbers

All of the above index number formulas and their relations to Translog and General-
ized Leontief functional forms were essentially calculated as a one stage aggregation
procedure. For example all inputs 1;


; N were aggregated directly into a single
input quantity index and a single input price index rather than into several quantity
98 8 Index Numbers and Flexible Functional Forms

and price subindexes. On the other hand, commodities typically are aggregated into
various subindexes for use in econometric models.
This leads to the following important question: are two stage aggregation pro-
duces (using index numbers formulas) exact or approximately exact? For example,
suppose that the Törnqvist quantity index number formula (8.6) and the correspond-
ing implicit price index formula are applied separately to two subsets of inputs (NA
and NB inputs, respectively, where NA C NB D N , the total number of inputs),
resulting in two quantity indexes QA ; QB
 A
X1
QA  log
X0A
NA
! !
A A A A A
1 X wi;1 xi;1 wi;0 xi;0 xi;1
D
PNA A A C PNA A A
log A
2 xi;0
i D1 j D1 wj;1 xj;1 j D1 wj;0 xj;0
 B (8.41)
B X1
Q  log
X0B
NB
! !
B B B B B
1 X wi;1 xi;1 wi;0 xi;0 xi;1
D
PNB B B C PNB B B
log B
2 xi;0
i D1 j D1 wj;1 xj;1 j D1 wj;0 xj;0

and corresponding implicit prices indexes pQ A , pQ B . Then the same quantity index
number formula (8.6) is applied to the subindexes QA , QB , pQ A , pQ B , resulting in a
y
quantity index Q:
 
y  log X1
Q
2
X0
pQ1A Q1A pQ0A Q0A
   A
1 Q1
D C log (8.42)
2 pQ1A Q1A C pQ1B Q1B pQ0A Q0A C pQ0B Q0B Q0A
 B B B B   B
1 pQ1 Q1 pQ0 Q0 Q1
C A A B B
C A A B B
log
2 pQ1 Q1 C pQ1 Q1 pQ0 Q0 C pQ0 Q0 Q0B

Are the results of one stage and two stage aggregation identical? For particular,
is the quantity index Q y calculated in (8.41)–(8.42) exact for a constant returns
to scale Translog production function y D f .x 1 ; : : : ; x N /, i.e. does Q y equal
log.f .x1 /=f .x0 //?
In general the two stage aggregation procedure (8.41)–(8.42) is not exact, i.e. Q y¤
log.f .x1 /=f .x0 //. This result is not surprising, since results in the previous chapter
indicated that two stage budgeting is correct only under strong restrictions on the
structure of production or utility functions (e.g. homothetic weak separability).
On the other hand the two stage aggregation procedure (8.41)–(8.42) is approx-
imately exact, i.e. Q y approximates log .f .x1 /=f .x0 //. Moreover, two stage aggre-
gation procedures based on known superlative index number formulas are approxi-
mately exact. This can be explained very briefly as follows (see Diewert 1978).
8.4 Two-stage aggregation with superlative index numbers 99

The Vartia I price index P V and quantity index QV , where

N  
V
X pi;1
log P  si log
pi;0
iD1
N  
V
X xi;1
log Q  si log
xi;0
i D1
2 , PN PN 3
pi;1 xi;1 pi;0 xi;0 p x
j D1 j;1 j;1 p x
j D1 j;0 j;0
where si  4 P .P 5 ;
log .pi;1 xi;1 =pi;0 xi;0 / log N
p x N
p x
j D1 j;1 j;1 j D1 j;0 j;0
(8.43)
are know to be consistent in aggregation, i.e. one stage and two stage aggregation
procedures based on (8.43) lead to identical index numbers. Unfortunately these
Vartia indexes are exact for a constant returns to scale Cobb-Douglas production
function y D f .x1 ; : : : ; xN / rather then for a second order flexible functional
form.4
Nevertheless it can be shown that the vartia quantity index QV differentially
approximates the Törnqvist quantity. index QT  .X1 =X0 / (8.6) to the second
order at any point where the prices and quantities are the same for the two time
periods comparison .p1 D p0 ; x1 D x0 /, i.e.

QV .z/ D QT .z/
@QV .z/ @QT .z/
D
@´i @´i (8.44)
@2 QV .z/ @2 QT .z/
D for all i; j
@´i @´j @´i @´j

where z  .p1 ; p0 ; x1 ; x2 / and p1 D p0 , x1 D x0 . Identical results hold for

the Vartia price index P V and the Törnqvist price index p T  .W1 =W0 / (8.18).
These results do not require any assumptions about optimizing behavior log agents.
Moreover, these results can be extended to other superlative indexes, e.g. indexes
corresponding to Generalized Leontief production functions or cost functions.
Therefore two stage aggregation procedures using superlative index number for-
mulas are approximately correct (exact) for relatively small changes in prices and
quantities between the comparison time periods t D 0; 1. These changes in prices
and quantities are usually smaller when indexes are constructed by chaining obser-
vations in successive period rather than using a constant base period. On this basis
it is recommended that a Törnqvist quantity macroindex or subindex (8.6), e.g., be
constructed as

4
Laspeyres and paasche index numbers are also consistent in aggregation, but the corresponding
production are much more restrictive than the Cobb-Douglas (see Section 8.1).
100 8 Index Numbers and Flexible Functional Forms
!
A
XNA xi;t
QA
t D sA log t D 1;


; T (8.45)
i D1 i;t A
xi;t 1
!
A A A A
A 1 wi;t xi;t wi;t 1 xi;t 1
where si;t D PNA C PN
2 A A
wj;t xj;t A A
wj;t A
j D1 j D1 1 xj;t 1
(chaining observations in successive periods) rather than as
!
A
A
XNA
A
xi;t
Qt D s log t D 1;


; T (8.46)
i D1 i;t A
xi;0
!
A A A A
A 1 w i;t x i;t w i;0 x i;0
where si;t D PNA A A C PNA A A
2
j D1 wj;t xj;t j D1 wj;0 xj;0
(using a constant base period t D 0).

8.5 Conclusion

The above results indicate that, at least in the case of simple static maximization
models, various index number procedures can be used to obtain approximately con-
sistent aggregation of commodities provided that the variation in prices and quan-
tities between comparison periods is small. For time series data this condition can
usually be satisfied by chaining observations in successive periods.
However in the case of cross-section data there may be substantial variation in
quantities or prices between successive periods, and here different superlative index
number formulas may lead to significantly different results. In this case it may be
useful to compare the variation in the N quantity ratios xi;1 =xi;0 to the variation in
the N price ratios pi;1 =pi;0 .
Usually there is much less variation in the price ratios than in the quantity
PNA A A A
ratios. Then a directly defined price index P tA D i D1 si;t log.pi;t =pi;t 1 / is
less sensitive to the variation in data than is a directly defined quantity index
PNA A A A
QA t D i D1 si;t log.xi;t =xi;t 1 / (since both equations use the same shares s t ).
A

This suggests that the best strategy in this case is to calculate the price indexes di-
rectly and to employ the corresponding implicit quantity indexes (see Allen 1981,
pp. 430-435).

References

1. Allen, R. C. and Diewert, W. E. (1981). Direct versus implicit superlative index number for-
mulae. The Review of Economics and Statistics, 63(3):430–435.
2. Deaton, A. and Muellbauer, J. (1980). Economics and Consumer Behavior. Cambridge Uni-
versity Press.
3. Diewert, W. E. (1976). The Economic Theory of Index Number: A Survey, volume 1. Elsevier
Science Publishers.
References 101

4. Diewert, W. E. (1978). Superlative index numbers and consistency in aggregation. Economet-

rica, 46(4):883–900.
5. Fuss, M., McFadden, D., and Mundlak, Y. (1978). A Survey of Functional Forms in the Eco-
nomic Analysis of Production, volume 1. McMaster University Archive for the History of
Economic Thought.
6. Hulten, C. R. (1973). Divisia index numbers. Econometrica, 41(6):1017–1025.
102 8 Index Numbers and Flexible Functional Forms

8.6 Laspeyres and Paasche cost of living indexes

)
min px
x
! E.p; u/
s.t. u.x/ D u
u.x/ homothetic E.p; u/ D u E.p; 1/.
„ ƒ‚ …
e.p/
True cost of living (COL) index conditional on u :

P1 E.p1 ; u /
D :
P2 E.p0 ; u /

Homotheticity ) COL index is independent of any u :

P1 u
e.p1 / e.p1 /
D  D :
P0 u
e.p0 / e.p0 /

Laspeyres COL index is > true COL (at u0 ):

L
x0 p1 > E.p1 ; u0 /

P1
 :
P0 x0 p0 D E.p0 ; u0 /
„ ƒ‚ … „ ƒ‚ …
Laspeyres COL index true COL index

Paasche COL index is < true COL(at u1 ):

P
x1 p1 D E.p1 ; u1 /

P1
 :
P0 x p0 > E.p0 ; u1 /
„ ƒ‚ 1 … „ ƒ‚ …
Paasche COL index true COL index

So (assuming homotheticity)
 P  L
P1 P1
< true COL < (8.47)
P0 P0

Note: (8.47) marks no assumptions about form of preferences (such as Translog),

except for homotheticity.
If (8.47) does no place close enough bounds on true COL, then we should specify
(e.g.) a Törnqvist consumer price index (assuming a Translog CRTS cost function
E.p; u/).
8.7 Fisher indexes (for inputs) 103

8.7 Fisher indexes (for inputs)

CRT S ! C.w; y/ D yC.w; 1/ where C.w; 1/  c.w/: unit cost function.

Assume the following quadratic c.w/ function:
hX X i1=2
c.w/ D aij wi wj (8.48)
i j

Define the following price index:

 F " L  P # 12
W1 W1 W1
D (8.49)
W0 W0 W0

This is called a Fisher price index (it is a geometric mean of a Laspeyres and
Paasche index).
We can prove the following result:

Theorem 8.1. Assume CRTS and the quadratic cost function (8.48) (all inputs are
at static cost minimizing equilibrium). Then
 F
W1 c.w1 / AC1
D  :
W0 c.w0 / AC0

Alternatively define a Fisher input quantity index as

 F
X1 1
 .X1 =X0 /L .X1 =X0 /P 2

(8.50)
X0 „ ƒ‚ … „ ƒ‚ …
w0 x1 =w0 x0 w1 x1 =w1 x0

and assume a quadratic CRTS production function

hX X i1=2
yD bij xi xj (8.51)
i j

Then, assuming all inputs are at static cost minimizing equilibrium, we can show
 F
X1 y1
D :
X0 y0

Note: A quadratic cost function (8.48) is essentially equivalent to a quadratic pro-

duction function (8.51). (In contrast, Translog and G.L. forms are not “self-dual”).
104 8 Index Numbers and Flexible Functional Forms

Proof. Assuming CRTS, define the unit cost function c.w/ D C.w; y/=y. Assume
hX X i1=2
c.w/ D aij wi wj (8.52)
i j

So
@c.w/ 1 hX X i 1=2 X
D aij wi wj 2 aij wj .aij D aj i /
@wi 2 i j j
P (8.53)
j aij wj
D by (8.52)
c.w/
So P
@c.w/=@wi j aij wj
D (8.54)
c.w/ c.w/2
Cost minimization for all inputs implies Shephard’s Lemma

@c.w/
y D xi : (8.55)
@wi
P
Dividing (8.55) by yc.w/  i wi xi ,

xi @c.w/=@wi
P D : (8.56)
i wi xi c.w/

The Fisher input price index is

 F
W1 h i1=2
D .W1 =W0 /L .W1 =W0 /P
W0
D Œ.w1 x0 =w0 x0 /
.w1 x1 =w0 x1 /1=2
D Œ.w1 x0 =w0 x0 /=.w0 x1 =w1 x1 /1=2
@c.w1 /=@w 1=2
  
@c.w0 /=@w
D w1 w0 by (8.56)
c.w0 / c.w1 /
aij wj;1 1=2
X X  
aij wj;0 X X
D wi;1 w i;0 by (8.54)
i j c.w0 /2 i j c.w1 /2
h . i1=2
D 1=c.w0 /2 1=c.w1 /2
D c.w1 /=c.w0 /
Chapter 9
Measuring Technical Change

In previous chapters we assumed the existence of an unchanged production function

y D f .x/ as transformation function g.y; x/ D 0, where output levels y are deter-
mined by input levels x. In contrast here we allow for shifts in the technology of the
firm or industry that lead to changes in output levels y that cannot be accounted for
solely by changes in input levels x.
In principle studies of productivity as in this chapter control for changes in the
quality of measured inputs. Consequently application of these methods to a partic-
ular sector of the economy (e.g. agriculture) will not (in principle) quantify the ef-
fect of improved quality of measured inputs on the sector’s output. Improvement in
quality of inputs should be analyzed as technical changes in the industries producing
those factors. In order to study the effect of improvements in quality of measured in-
puts as well as changes in levels of unmeasured inputs on agricultural output, it has
been suggested that agriculture and manufacturing industries supplying measured
inputs to agriculture should be combined into1 a single sector where these inputs
are treated as intermediate goods (Kisler and Peterson 1981).
This chapter introduces several approaches to the measurement of technical
change. Section 9.1 considers econometric production models incorporating a time
tread as a proxy for technical change. Section 9.2 summarizes index number pro-
cedures for calculating changes in productivity without recourse to econometrics
or specification of specific functional forms for the production function, cost func-
tion or profit function (in this sense the procedures are non-parametric). Section 9.3
summarizes parametric index number procedures for calculating changes in produc-
tivity without recourse to econometrics. These sections 9.2 and 9.3 both assume that
there is no variation in utilization rates of capital over time; i.e. the industry or firm
is assumed to be in long-run equilibrium. Section 9.4 demonstrates that (under re-

1
In principle it is possible to incorporate technical change into the production function by expand-
ing the list of inputs x so as to include all variables that contribute to technical change, e.g. research
results of agricultural experiment stations and agricultural extension activities. However these ad-
ditional inputs are not easily quantified. This is the primary rationalization for treating technical
changes as a residual of output changes that is unexplained by changes in levels of observable
inputs.

105
106 9 Measuring Technical Change

strictive assumptions) these non-econometric procedures can be modified to allow

for variable utilization rates in capital, and that econometric procedures can (under
certain restrictive assumptions) allow for variable utilization rates in capital. Section
9.5 conclude the chapter.

9.1 Dual cost and profit functions with technical change

In both primal and dual econometric models of the industry or firm, technical
changes usually is proxied simply by a time trend variable t D 1; 2; 3;


; T where
T is the number of time periods. By allowing for interactions between this time trend
and other variables in the model, this specification is designed to proxy more than a
regular secular time trend for technical changes over historical time. Of course the
time trend may also proxy secular trends in any relevant variables that have been ex-
cluded from the model. In addition it is assumed that technical change is exogenous
to the industry or firm.
A multiple output Translog cost function C D C.w; y/ can be generalized to
incorporate a time trend t D 1; 2; 3;


; as follows:
N
X M
X
log C D ˛0 C ˛i log wi C ˇi log yi C '1 t
i D1 i D1
N N
1 XX
C ˛ij log wi log wj
2
i D1 j D1
M M
1 XX
C ˇij log yi log yj (9.1)
2
i D1 j D1
N M
1 XX
C ij log wi log yj
2
i D1 j D1
N
X M
X
C i t log wi C ıi t log yi C '2 t 2
iD1 i D1

Shephard’s lemma implies the following factor share equations:

N M
wi xi X X
si  D ˛i C ˛ij log wj C ij log yj C i t i D 1;


; N (9.2)
C
j D1 j D1

Assuming competitive profit maximization, the first order conditions pj D @C.w; y; t /=@yj
.j D 1;


; M / and (9.1) imply
9.1 Dual cost and profit functions with technical change 107
pj yj @ log C
D
C @ log yj
N
X M
X (9.3)
D ˇj C ij log wi C ˇj i log yi C ıj t j D 1;


; M
i D1 i D1

Differentiating (9.1) with respect to t ,

@ log C @C.w; y; t /=C.w; y; t /


@t @t
N
X XM (9.4)
D '1 C i log wi C ıj log yj C 2'2 t
i D1 j D1

Assuming technical progress (not regress), @ log C =@t < 0. Note that equation (9.4)
cannot be estimated directly because the percentage reduction in cost due to techni-
cal change (@ log C =@t ) is unobserved, and also note that the coefficients '1 , '2 of
(9.4) cannot be inferred from estimates of factor demand and output supply equation
(9.2)–(9.3). In order to obtain estimates of all coefficients of (9.4), it is necessary to
estimate directly equation (9.1) defining the Translog cost function. Unfortunately
the data will often permit estimation of equation (9.2)–(9.3), but not direct estima-
tion of the cost function equation (9.1).
Nevertheless we can easily derive a measure of technical change @C.w; y; t /=@t
directly from estimates of the factor share equation (9.2). Since C.w; y; t / 
PN
iD1 xi .w; y; t /wi ,

N
@C.w; y; t / X @xi .w; y; t /
 wi (9.5)
@t @t
i D1

Differentiating si .w; y; t /  wi xi .w; y; t /=C.w; y; t / with respect to t and substi-

@s
tuting in (9.5) yields @t D A @x
@t
where matrix A has full rank.2 Then @x @t
can be
@x 1 @s
calculated as @t D A  using the estimates of @t from the share equations (9.2),
and in turn @C
@t
can be calculated using (9.5).
A single output generalized Leontief cost function with technical change can be
specified as
XX p p X X
C Dy aij wi wj C y 2 ai wi C ty bi wi (9.6)
i j i i

2
w1
1 wC1 w1 w2 w1 wN






2 3
C C C
C C
w1 w2 w2
6 C C C
1 wC2


w2 wN
C C
7
AD6
6
:: :: :: ::
7
7
4 : : : : 5
w1 wN w2 wN wN wN
C C C C



C
.1 C
/
which generally has full rank. Deleting the equation for share sN from the estimation of (9.2),
PN 1
N  1 i D1 i ,
108 9 Measuring Technical Change

Here @C.w; y; t /=@t D y N

P
i D1 bi wi , and all coefficients of this equation can be
recovered directly from the estimated factor demand equations (in contrast to the
Translog).
The shift in the underlying production function or transformation function due to
technical change can easily be calculated from the reduction in cost @C.w; y; t /=@y.
Assuming competitive cost minimization and a production function y D f .x; t / for
a single output y, the increase in output @f .x ; t /=@t due to technical change can be
calculated as follows:
@f .x ; t / @C.w; y; t / @C.w; y; t /
D = (9.7)
@t @t @y
or, assuming competitive profit maximization,

@f .x  ; t / @C.w; y; t /
D =p (9.8)
@t @t
In the case of constant returns to scale in production C.w; y; t / D Cy .w; y; t /y
(Euler’s theorem), and in turn (9.7) implies

@f .x ; t / @C.w; y; t /
=y D =C (9.9)
@t @t
Thus under constant returns to scale the rate of growth of output (at equilibrium x )
is equal to the rate of reduction in cost due to technical change. Similarly in the case
of a multiple output transformation function y1 D g.y2 ;


; yM ; x; t /  y1 .Qy; x; t /
and assuming competitive profit maximization, the shift in transformation function
due to technical change @g.Qy; x; t /=@t can be calculated as

@g.Qy; x; t / @C.w; y; t /
D =p1 (9.10)
@t @t
These relations (9.7)-(9.10) are proved in the next section 9.2.
Biases as well as magnitudes of technical change can be calculated from cost
functions. For this purpose it is convenient to define Hicks-neutral technical change
as technical change that does not alter the firm or industry’s input expansion path.
Then the change in cost shares @si .w; y; t /=@t .i D 1;


; N / provides a measure
of bias if the production function is homothetic, i.e. the expansion in input space is
linear from the origin for a given state of technology.
However the change in shares @s.w; y; t /=@t does not provide an accurate mea-
sure of biases in technical change when the production function is non-homothetic.
For example suppose that Hicks-neutral technical change occurs, i.e. technical
change does not alter the input expansion path. This technical change simply leads
to a re-numbering of the output levels for given isoquants (@f .x ; t /=@t > 0 and
output y constant implies the firm moves to a lower isoquant). If the expansion path
is not linear and the firm’s output level must be held constant as this reindexing oc-
curs, then the firm moves to a different isoquant with a different cost-minimizing
9.1 Dual cost and profit functions with technical change 109

ratio of inputs. Thus the changes in shares @si .w; y; t /=@t .i D 1;


; N / are not
equal to zero even though technical change is Hicks-neutral.
In order to correct for non-homotheticity in calculating biases in technical
change, note that the change in shares can be decomposed into a scale effect due
to movement along the initial expansion path and a bias effect due to the shift in the
expansion path. The share equation si D si .w; y; t / can be defined equivalently as
si D sOi .w; i  .t /; t / where i  .t / indexes the isoquant yielding output level y given
technology t. Differentiation of this composite function with respect to t yields

@si .w; y; t / @Osi .w; i  ; t / @si .w; y; t / @f .x ; t /

D C (9.11)
@t @t @y @t

where @Osi .w; i  ; t /=@t provides a true measure of bias in technical change. Combin-
ing (9.7) and (9.11), a true measure of bias in technical change for a non-homothetic
production function can be defined in terms of a cost function as

@si .w; y; t / @si .w; y; t / @C.w; y; t / @C.w; y; t /

C = i D 1;


; N (9.12)
@t @y @t @y
or in elasticity terms as

@ log si .w; y; t / @ log si .w; y; t / @ log C.w; y; t / @ log C.w; y; t /

C = i D 1;


; N
@t @ log y @t @ log y
(9.13)
@ log si .w;y;t /
Homotheticity implies @si .w;y;t
@y
/
D @ log y
D 0 .i D 1;


; N /. Similarly
in the case of multiple outputs, a true measure of bias in technical change can be
calculated as
M
@ log si .w; y; t / X @ log s.w; y; t / @ log C.w; y; t / @ log C.w; y; t /
C = (9.14)
@t @ log yj @t @ log yj
j D1

Technical change can be incorporated into dual profit functions in a manner anal-
ogous to the above treatment of cost functions. Here we simply note the following
relation between changes in profits and costs due to technical change

@.w; p; t /=@t D @C.w; y  ; t /=@t (9.15)

This relation is proved in the next section 9.2.

Finally it is important to note that most of the above discussion applies equally
well to short-run and long-run equilibrium models. The above models assumed that
all inputs are freely variable and attain static equilibrium levels. In contrast we could
postulate, e.g. , a Translog cost function C.w; y; k; t / conditional on the stocks K
of quasi-fixed inputs.
In place of (9.7) we have the following relation between the shift @f .x  ; K; t /=@t
in the production function and the shift @C.w; y; k; t /=@t in cost:
110 9 Measuring Technical Change

@f .x  ; K; t / @C.w; y; K; t / @C.w; y; K; t /
D = (9.16)
@t @t @y

where Cy .w; y; K; t / D p assuming competitive profit maximization. On the other

hand the measurement of biases is complicated somewhat when the cost function is
short-run equilibrium in nature.

9.2 Non-Parametric index number calculations of changes in

productivity

In this section we consider methods for calculating technical change that require nei-
ther econometrics nor the specification of particular functional forms for a produc-
tion function as cost function. However this index number approach to productivity
is defined in terms of continuous time, and errors in approximation result from the
use of discrete time data. Moreover these calculations usually assume that all inputs
are freely variable and are at static long-run equilibrium levels.
Initially assume a single output production function y D f .x; t /. Then output
at time t is related to inputs at time t and technology as indicated by the relation
y.t / D f .x.t /; t /. Differentiating with respect to t yields
N
X @f .x; t / @f .x; t /
yV D xV i C (9.17)
@x i @t
i D1

where yV  @y.t@t
/
, xV  @x.t
@t
/
denotes the change in output and input levels with
respect to time. The standard first order conditions for static competitive cost min-
 ;t /
imization are w i Cy .w; y; t / @f .x@x i
D 0 .i D 1;


; N /, and substituting these
into (9.17) yields
N 
wi

X @f .x; t /
yV D xV i C (9.18)
Cy .w; y; t / @t
i D1

Competitive profit maximization implies Cy .w; y; t / D p, and substituting this

into (9.18) yields
N  i
X w @f .x; t /
yV D xV i C (9.19)
p @t
i D1

or equivalently
N  i i i
X w x xV @f .x; t /
V D
y=y i
C =y (9.20)
p y x @t
iD1

The weightings w i x i =py for input changes xV i =x i define a Divisia quantity index
for inputs in continuous time. Thus, assuming continuous time and all inputs are at
9.2 Non-Parametric index number calculations of changes in productivity 111

static long-run equilibrium levels, technical change @f .x; t /=@t could be calculated
as the residual in (9.18) or (9.19)-(9.20).
However data is measured at discrete time intervals rather than continuously, and
equation (9.19) or (9.20) can only be approximated using discrete data. For example
integrating (9.19) over an interval t D 0; 1 yields
N Z 1 1
w i @x i .t / @f .x.t /; t /
X Z
y.1/ y.0/ D .t / dt C dt (9.21)
t D0 p @t tD0 @t
i D1

However, unless all price ratios w i =p are independent of t .i D 1;


; N /,

R1 i i
closed form solutions for the integrals tD0 wp .t / @x@t.t / dt generally cannot be de-
fined. (9.20) is most commonly approximated using Törnqvist approximations to
the Divisia index:
N
1 wi xi wi xi x i .t / x i .t
 
y.t / y.t 1/ X 1/
TF  .t / C .t 1/
y.t / 2 py py x i .t /
i D1
(9.22)
Rt @f .x.v/;v/
where Tf  vDt 1 @v
=y.v/dv denotes the integral of the residual in (9.19)
attributed to technical change.
There is a further complication in interpreting the calculated shifts in the pro-
duction function @f .x  ; t /=@t. The equilibrium point x  where the continuous on
discrete calculations are made varies over time. Therefore, in order to interpret these
results as simple comparable indicators of shifts in the production function over
time, it is necessary to assume Hicks-neutral technical change and constant returns
to scale. Similar comments apply to the interpretation of other index number calcu-
lations of productivity presented in this section and the following section C.
In the case of multiple outputs y D .y 1 ;


; y M /, we can define a transfor-
mation function normalized on y 1 : y 1 D g.y; Q x; t / where yQ  .y 2 ;


; y M /.
1
Differentiating y .t / D g.y.tQ /; x.t /; t / with respect to t yields
M
X N
X
yV 1 D Q x; t /yVQ i C
gyQ .y; Q x; t /xV i C @g.y;
gx i .y; Q x; t /=@t (9.23)
iD2 i D1

The competitive maximization problem maxy;x; L  py wx .g.y; Q x; t /

y 1 / has first order conditions p i Cp 1 gyi .y;
Q x; t / D 0 .i D 2;


; M /, w i p i gx i .y;
Q x; t / D
0 .i D 1;


; N /, and substituting these into (9.23) yields
M
X N
X
i 1 i
.p =p /yV D .w i =p 1 /xV i C @g.y;
Q x; t /=@t (9.24)
i D1 i D1

However the numerical measure of technical change @g.y; Q x; t /=@t varies with
the choice of output i , i.e. the measure of technical change is not symmetric with
respect to the normalization of the transformation function.
112 9 Measuring Technical Change

In the case of multiple outputs a move useful measure of technical change may
be obtained in terms of the cost function C.w; y; t /. At time t C.w.t /; y.t /; t / 
PN i i
iD1 w .t /x .t /, and differentiating with respect to t yields

N N N
X X @C.w; y; t / X X
Cw i .w; y; t /wV i C Cyi .w; y; t /yV i C D wV i x i C w i xV i
@t
i D1 j i D1 iD1
(9.25)
Competitive cost minimization implies Cw i .w; y; t / D x i .w; y; t / .i D 2;


; N /
(Shephard’s lemma), and substituting these conditions into (9.25) yields
N M
@C.w; y; t / X X
D w i xV i Cyj .w; y; t /yV j (9.26)
@t
iD1 j D1

Competitive profit maximization further implies Cyj .w; y; t / D p j .j D

1;


; M /, and substituting into (9.26) yields
N
X M
X
@C.w; y; t /=@t D w i xV i p i yV j (9.27)
iD1 j D1

or equivalently
N  i i i M  j
p yj yV j

@C.w; y; t / X w x xV X
=C D (9.28)
@t C xi C yj
i D1 j D1

How the simple relations between primal measures and dual cost and profit mea-
sures of technical change can easily be established as follows. Comparing (9.18)
and (9.26) for the case of a single output and cost minimization establishes

@C.w; y; t / @f .x  ; t / @C.w; y; t /
D (9.29)
@t @t @y
or in the case of competitive profit maximization

@C.w; y; t / @f .x  ; t /
D p (9.30)
@t @t
Similarly comparing (9.24) and (9.27) in the case of multiple outputs and profit
maximization establishes
@C.w; y; t / @g.yQ  ; x  ; t /
D p1 (9.31)
@t @t
In term of the profit function .w; p; t / we have the identity .w.t /; p.t /; t / 
p.t /y.t / w.t /x.t / for profits at time t, and differentiating with respect to t and
applying Hotelling’s lemma yields
9.3 Parametric index number calculations of changes in productivity 113

M N
@.w; p; t / X X
D p i yV i w i xV i (9.32)
@t
i D1 i D1

Comparing (9.27) and (9.32) establishes

@C.w; y  ; t / @.w; p; t /
D (9.33)
@t @t

9.3 Parametric index number calculations of changes in

productivity

The Divisia approaches of the previous section required the approximation of con-
tinuous time derivatives by discrete differences but did not require the specification
of a particular functional form for a production function or cost function. In this
section we discuss an alternative approach to calculating changes in productivity: a
flexible functional form is assumed for a production function or cost function, and
then an index number formula is derived that is consistent with the functional form
and with discrete time data. Here there are no errors in approximation due to the use
of data for discrete time intervals, but there are errors in approximation to the true
functional form for the production on cost function. Here we derive index numbers
formulas for technical change corresponding to thus different functional forms: a
Translog cost function, a Translog transformation function, and a Translog profit
function.
First assume a multiple output Translog cost function C.w; y; t / as in (9.1). The
underlying transformation function need not be constant returns to scale. Since this
cost function is a quadratic form, the quadratic identity 8.11 of chapter 8 implies
114 9 Measuring Technical Change
N  
1 X @ log C1 @ log C0
log w1i =w0i


log .C1 =C0 / D i
C i
2 @ log w @ log w
iD1
M  
1 X @ log C1 @ log C0
log y1i =y0i


C C
2 @ log yi @ log yi
i D1
 
1 @ log C1 @ log C0
C C .t1 t0 /
2 @t @t
N 
1 X @C1 w1i @C0 w0i

log w1i =w0i


D i
C i
2 @w C1 @w C0
iD1
M 
1 X @C1 y1i @C0 y0i

C C log y1i =y0i

(9.34)
2 @yi C1 @yi C0
i D1
 
1 @C1 @C0
C =C1 C =C0 since t1 D t0 C 1
2 @t @t
N 
1 X w1i x1i wi xi

C 0 0 log w1i =w0i


D
2 C1 C0
iD1
M 
1 X p1i y1i pi y i

C 0 0 log y1i =y0i


C
2 C1 C0
i D1
 
1 @C1 @C0
C =C1 C =C0
2 @t @t

by Shephard’s lemma and @C =@yi D p i .i D 1;


; M /. This index number equa-

tion is identical to 8.26 discussed on pages 8.10-8.11 of chapter 8, except that a time
trend variable t D 1; 2; 3;


is incorporated into the Translog cost function. The
firm is assumed to be a competitive profit maximizer with all inputs variable and
attaining static long-run equilibrium levels.
Rearranging (9.34)
N 
(  i )
1 X w1i x1i w0i x0i

1 C C
 w1
T C T0 D log .C1 =C0 / C log
2 1 2 C1 C0 w0i
iD1
(9.35)
M 
1 X p1i y1i p0i y0i
  i
y1
C log
2
iD1
C1 C0 y0i
PN
where TVC  @C.wV@t;yV ;tV / =C.wV ; yV ; tV /, CV  i i
iD1 wV xV .V D 0; 1/ .
V
1 C C
2
.T1 C T0 / is the average percentage reduction in cost due to technical change at
time t D 0; 1.
The second set of terms within brackets f


g on the right hand side of (9.35) is
the logarithm of a Törnqvist price index .W1 =W0 / for inputs (seeequation
A 8.18), so
the brackets f


g enclose an implicit Törnqvist quantity index x1 =x0 for input
9.3 Parametric index number calculations of changes in productivity 115

A 
x1 =x0 D .C1 =C0 / = .w1 =w0 / by the factor reversal equation analogous to 8.16.
The term on the right hand side of (9.35) that is outside the brackets can be inter-
preted as the logarithm of a quantity index .Y1 =Y0 / for outputs (constant returns to
scale in production would imply C D py and this index would be equivalent to
a Törnqvist quantity index for outputs). Hence the average percentage reduction in
cost is the logarithm of the ratio of input and output quantity indexes:
1 C
2
 h
A
T1 C T0C D log .x1 =x0 /=.Y1 =Y0 /
i
(9.36)

This calculated change in cost can easily be related to shifts in the production or
transformation function at equilibrium levels of commodities x  , y  using equations
(9.7)-(9.10).
Second, assume a constant returns to scale Translog production function y D
f .x; t / and competitive cost minimization with all inputs variable at long-run equi-
librium levels. Proceeding as in the proof of 8.8, we obtain
N 
1 X w1i x1i w0i x0i
    i
1 t t

y1 x1
T C T0 D log C log (9.37)
2 1 y0 2 w1 x1 w0 x0 x0i
i D1

where TVt  @f .x@tV ;tV /

=yV .v D 0; 1/. Here the average percentage change in pro-
ductivity is equal to the logarithm of the ratio of output and input quantity indexes:
1 t h i
T1 C T0t D log .y1 =y0 /=.x1 =x0 /T


(9.38)
2
where .x1 =x0 /T is the Törnqvist quantity index for inputs. Alternatively if we as-
sume a Translog production function (which is not necessarily constant returns to
scale) and competitive profit maximization, then we obtain the closely related index
number formula
N 
1 X w1i x1i w0i x0i
    i
1 t t

y1 x1
T1 C T0 D log C log (9.39)
2 y0 2
i D1
p1 y1 p0 y 0 x0i

Similarly in the case of multiple outputs assume a Translog transformation func-

tion y 1 D g.y;Q x; t / .yQ  .y 2 ;


; y M // and competitive profit maximizing be-
havior with all inputs variable. Then
M 
1 X p1i y1i p0i y0i
    i
1 g g
y1 y1
T C T0 D log C log
2 1 y0 2 p1 y1 p0 y0 y0i
i D1
(9.40)
N 
1 X wi xi w0i x0i
  i
1 1 x1
C log
2
i D1
p1 y1 p0 y0 x0i
116 9 Measuring Technical Change

@g.yQ  ;x  ;t /
where TVg  V
@t
V V
=yV .V D 0; 1/. This measure of technical change varies
in a simple manner with the commodity chosen as numeraire in the transformation
function. For example, given the alternative normalization y 1 D g.y 2 ;


; y M ; x; t /
and y 2 D h.y 1 ; y 3 ; y 4 ;


; y M ; x; t /,

@g.y 2 ; y M  ; x  ; t /=@t D .p 2 =p 1 /@h.y 1 ; y 3 ;


; y M  ; x  ; t /=@t (9.41)

Third, assume a Translog profit function and competitive profit maximization

with all inputs variable at long-run equilibrium levels. The quadratic identity and
Hotelling’s lemma establish
M 
1 X p1i y1i pi y i
    i
1  1 p1
T1 C T0 D log C 0 0 log


2 0 2 1
iD1
0 p0i
(9.42)
N 
w1i x1i wi xi
  i
1 X w1
C C 0 0 log
2
iD1
1 0 w0i

where TV  @.wV@t;pV ;tV / =.wV ; pV ; tV /. This measure of the effect of technical
change on profits .w; p; t / is easily related to changes in costs and shifts in the
production function or transformation function using equations (9.7)-(9.10). Inter-
preting the second and third terms of the right hand side of (9.42) as price indexes
for outputs and
1  

inputs, respectively, and noting that   py wx, it follows that
2
T 1 C T 0 is the logarithm of the ratio of implicit quantity indexes for outputs
and for inputs.
The behavioral models of the firm employed in sections B and C have assumed
static long-run equilibrium and lead to index number formulas such as (9.20) that
are defined in terms of the flows x t of all inputs used in production. Since the flow
of capital services is not generally observable, it is usually assumed that the flow of
capital services is proportional to the stock of capital assets (this assumption may
be reasonable at long-run equilibrium). Then the growth rate of the capital service
flow xV tk =x tk is equal to the growth rate of the capital stock KV t =K t .
The capital stock K t is often approximated by the perpetual inventory method:
K t D I t 1 C.1 ı/K t 1 where I t 1 denotes gross investment at time t 1 and ı is a
constant rate of depreciation, and substituting backwards for K t 1 ; K t 2 ;


; K t S
yields the approximation
S
X
Kt  It 1 C .1 ı/t SC1
It S (9.43)
SD2

Thus time series data on capital stock K and in turn the growth rate of the capital
service flow is approximated from data on gross investment I and an assumed rate
of depreciation ı. Then the rate of technical change is calculated using index number
formulas such as (9.20), (9.28), (9.35), (9.37) (e.g. Ball 1985).
9.4 Incorporating variable utilization rates for capital 117

9.4 Incorporating variable utilization rates for capital

There is considerable evidence that utilization rates of capital vary significantly over
time. This implies that firms generally are not in long-run equilibrium, and in turn
that (a) flows of capital services are not in fixed proportion to the levels of capital
stocks and (b) the marginal value product of capital services is not equal to a market
wage or rental rate. Nevertheless the index number procedures of sections B-C and
many econometric models incorporate these assumptions.
One approach to avoiding or reducing these problems in econometric models
stems from the assumption that the rate of depreciation for capital varies with the
utilization rate of capital. Then the firm’s short-run profit maximization problem can
be written as
f
max p f .x;K t ;K t /
wx t C pQ K K tf  .w; p; pQk ; K t / (9.44)
f
.x t ;K t /

where K t  capital stock predetermined at beginning of period t, K tf  position of

K t remaining at end of period t, and pQ K  p K =.1 C r/ is the asset price of capital
at the end of period t discounted back to the beginning of period t . The derivatives
of the production function are fx ./P > 0, fK ./P > 0 and f t f ./P < 0 (an increase
in K tf for a given initial stock K t implies a lower depreciation rate and utilization
rate of the initial stock K t ). However there is a serious difficulty in implement-
ing models such as (9.44) where the depreciation rate as well as utilization rate of
capital is treated as endogenous: time series data on capital stocks have invariably
been constructed assuming a constant rate of depreciation, and there are substantial
econometric problems in the estimation of (9.44) where K t is unobserved (see Ep-
stein and Denny, 1980 for an illustration). Hereafter we shall assume for simplicity
that variations in capital utilization do not influence the rate of depreciation.
Next consider the effects of short-run equilibrium on non-parametric Divisia in-
dex number calculations of technical change, as discussed in section B. Let the pro-
duction function be y D f .x; K; t / where x denotes flows of variable inputs and K
denotes stocks of predetermined (quasi-fixed) capital inputs, and assume short-run
competitive profit maximization

max pf .x; K; t / wx  .w; p; K; t / (9.45)

x

Differentiating the identity y.t /  f .x.t /; K.t /; t / and applying the first order
conditions pfx i .x  ; K; t / w i D 0 .i D 1;


; N / for a solution to (9.45) yields
N  i i i
@f .x  ; K; t / X w x xV X @f .x  ; K; t /
=y D y=y
V KV j =y (9.46)
@t py xi @K j
iD1 j

However the marginal products of capital @f .x  ; K; t /=@K j are not observed

j
directly are equal to price ratios w k =p only in a static long-run equilibrium.
118 9 Measuring Technical Change

Also note that the productivity index (9.46) employs the change in capital stocks
KV rather than a change in flow of capital services from the stocks. Equation (9.46)
provides an exact measure of change in productivity in continuous time. Thus errors

in calculations of changes in productivity @f .x@t;K;t / under the erroneous assumption
of long-run equilibrium are due to mismeasurement of weights for changes in capital
stocks kV rather than to mismeasurement of the flow of capital services (the correct
weights for KV are the unobserved marginal products of capital). In other words,
in productivity studies there is no need to derive capital service flows from capital
stocks irrespective of the industry being in long-run or short-run equilibrium. This
is an important result since the concept of capital service flows, which has been
widely employed in earlier productivity studies, is an artificial construct with little
empirical basis.

In the special case of constant returns to scale f .x; K; t / D f .x; K; t / and
one capital good K, the marginal product of capital can be calculated directly as
N  i
" #
@f .x  ; K; t / X w
D y x i =K (9.47)
@K p
i D1

Using Euler’s theorem and first order conditions for (9.45). Then changes in pro-
ductivity can be calculated directly from a discrete approximation to (9.46).
Similarly the variable cost function C D C.w; y; K; t /  wx and short-run
competitive behavior (9.45) imply
N  i i i X
M  i
@C.w; y  ; K; t / p yi yV i X @C.w; y  ; K; t / V i

X w x xV
=C D K =C
@t C xi C yi @K i
iD1 j D1 i
(9.48)
In the case of constant returns to scale in production and a single capital good K,
the unobserved shadow price of capital can be calculated as
0 1
M
@C.w; y; K; t / X
D @C p j yj A =K < 0 (9.49)
@K
j D1

(using C.w; y; K; t / D C.w; y; K; t / and Euler’s theorem).

Substituting (9.49) into (9.48), the resulting index number formula for technical
change can be approximated using data for discrete time intervals. This calculation
of technical change under constant returns to scale and a single capital good K
requires neither econometrics nor the assumption of long-run equilibrium.
Next consider the effects of short-run equilibrium on parametric Divisia index
number calculations of technical change, as discussed in section C. Assuming a
Translog variable cost function C.w; y; K; t / and competitive short-run profit max-
imization ((9.45)), equation (9.35) must be rewritten as
9.4 Incorporating variable utilization rates for capital 119
N 
( )
1 X w1i x1i wi xi
  i
1 C  w1
T1 C T0C D log .C1 =C0 / C 0 0 log
2 2
iD1
C1 C0 w0i
M 
p1i y1i pi y i
  i
1 X y1
C 0 0 log (9.50)
2
i D1
C1 C0 y0i
!
K1j
 
1 X @ log C1 @ log C0
C log
2
j
@ log Kj @ log Kj K0j

In general @@log
log C
Kj
 @K@C C
j = K j cannot be measured without recourse to econo-

metric estimation of the factor demand equations. However in the case of a single
quasi-fixed capital good K and constant returns to scale in production, @@ loglog C
K
can
be calculated from (9.49). Thus (as in non-parametric Divisia index number for-
mulas) given the assumptions of constant returns to scale and one quasi-fixed input,
parametric measures of technical change can be obtained without recourse to econo-
metrics or the assumption of long-run equilibrium.
Thus, if there is more than one quasi-fixed input or returns to scale are not con-
stant, econometric methods are required for the measurement of technical change.
For example we could postulate a short-run Translog cost function C.w; y; K; t /
analogous to (9.1) and estimate factor share equations for the variable inputs. Then
the change in technology @C.w;y;K;t
@t
/
can be calculated directly from the estimates
of the share equations as discussed in section A (page 9.4). Alternatively the shadow
prices can be calculated as
N
X @x i .w; y; K; t /
@C.w; y; K; t /
 wi (9.51)
@K j @K j
i D1

From the estimates of the share equations, and then the change in technology
1
T1C C T0C can be calculated from the index number equation (9.50). The disad-


2
vantage of this second approach is that equation (9.50) requires the assumption of
short-run competitive profit maximization, whereas estimation of the cost function
only requires the weaker assumption of short-run competitive cost minimization.
We conclude this section with a brief discussion of empirical measures of capac-
ity utilization based on microeconomic theory. Unexpected changes in output prices
or input prices are likely to lead to short-run combinations of variable and quasi-
fixed inputs that are inappropriate for the long-run, i.e. under or over-utilization of
capacity is likely in the short-run. A fruitful approach to the measurement of capac-
ity utilization is by comparing the observed level of output and “capacity output”.
One definition of capacity output is the output level corresponding to the minimum
point on the firm’s long-run average cost curve, but this definition is not useful due
to difficulties in identifying the long-run average cost curve.
A more useful definition of capacity output is the output level y C at which the
short-run average total cost curve (with quasi-fixed inputs fixed at their short-run
equilibrium levels) is tangent to the long-run average cost curve. If there is constant
120 9 Measuring Technical Change

returns to scale in the long-run, then capacity output corresponds to the minimum
point on the short-run average total cost curve (see diagram on following page). The
rate of capacity utilization C U is then defined as the ratio of actual output y to
capacity output y C :
C U  y=y C (9.52)
Econometric estimates of a short-run cost function C.w; y; K; t / can be em-
ployed in calculations of a capacity utilization index C U (Berrdt and Herse 1986).

9.5 Conclusion

In this chapter we have discussed several non-econometric and econometric ap-

proaches to the measurement of technical change. The non-econometric approaches
do not require a long time series of data or the use of highly aggregated data. On
the other hand these approaches cannot test hypothesis about technical change, and
these approaches can disentangle shifts in productivity and variations in capital uti-
lization rates only in the case of constant returns to scale and a single quasi-fixed
input. In contrast econometric approaches require a substantial number of obser-
vations and substantial aggregation of commodities, but these approaches can test
hypothesis and can calculate changes in productivity given multiple quasi-fixed in-
puts.
In practice non-econometric and econometric approaches to the measurement of
technical change can often be viewed as complementary. Most of the effort required
to construct non-econometric.
Measures of (full) Capacity Output
A.

A SRAC ≡ AC(w, wK , y, K) LRAC ≡ AC(w, wK , y)

yC C

Fig. 9.1 figure 1

9.5 Conclusion 121

B. Assuming constant
returns to scale in the production function y D f .x; K/,

f .x; K/ D f .x; k/ :

A SRAC0 ≡ AC(w, wK , y, K0 ) SRAC1 ≡ AC(w, wK , y, K1 )

y0C y1C C

Fig. 9.2 figure 2

Short-run average cost function (SRAC)

wx C w k K
AC.w; w K ; y; K/ D min
x @y
s.t. f .x; K/ D y

Long-run average cost function (LRAC):

wx C w k K
AC.w; w K ; y/ D min
x;K y
s.t. f .x; K/ D y

Index number calculations of technical change will also be required to construct

econometric models, and data problems may be indicated more clearly by non-
econometric measures of technical change. Thus the following sequence may be
appropriate: first construct non-econometric index number measures of technical
change, and then (data permitting) estimate a short-run equilibrium econometric
model.
However at least two serious problems remain for the econometric models and
Divisia index number formulas outlined here. First technical change is proxied by
a simple time trend t D 1; 2; 3;


(which may interact with other variables) or
is calculated as a simple residual. Thus there is no theory endogenizing technical
122 9 Measuring Technical Change

change to the model. In principle a well constructed theory of demand and supply
for changes in technology should improve both econometric and non-econometric
measures of technical change. Second, there has been little attempt to incorporate
explicitly into the analysis changes in quality of inputs supplied to the industry (sec
Berrdt 1983 and Tarr 1982).
4. It should be noted that cross-sectional differences in productivity (e.g. dif-
ferences in productivity between countries, regions or firms) can be measured by
parametric index number procedures somewhat similar to the procedures discussed
in section C (Caves, Christensen and Diewert 1982; Denny and Fuss 1983).
Index

C Gorman polar form 65

composite commodity 78 H
cost function
Hicksian consumer demands 28
short-run
Hotelling’s Lemma 12, 14
Generalized Leontief 59
normalized quadratic 59 I
Translog 59
cost of living indexes 102 implicitly separable 83
integrability problem 4
D in economics 32

Divisia index 90 L
dual cost function 1 Laspeyres index numbers 87
dual expenditure function 28 Laspeyres indexes 102
dual indirect utility function 27 Le Chatelier principle 17
dual profit function 11
dual profit function, restricted 18 M

E marginal frim 21
Marshallian consumer demands 28
equations of motion 24
N
Euler’s theorem 4
Normalized Quadratic indirect utility function
F 56
normalized quadratic cost function 55
factor reversal equation 92 normalized quadratic profit function 52
Fisher indexes 103 normalized quadratic short-run cost function
flexible functional forms 47 59
second order 49
Frobenius theorem 4 P

Paasche indexes 102

G
primal-dual relations 39
Generalized Leontief functional form 7 Q
Generalized Leontief short-run cost function
59 quadratic dual profit function 51

123
124 Index

quasi-homothetic 65 T

R Törnqvist index 90
Translog cost function 56
ratio of commodities 85 Translog dual profit function 53
ratio of marginal rates of substitution 85 Translog indirect utility function 57
Roy’s theorem 29 Translog short-run cost function 59
two-stage budgeting 79
S
V
second order flexible functional forms 49
separable
implicitly 83 Vartia I price index 99
weakly 80
Shephard’s Lemma 2 W
Slutsky equation 31
sub-utility function 80 Walras Law 63
superlative 90 weakly separable 80