Directory UMM :Data Elmu:jurnal:I:International Journal of Production Economics:Vol67.Issue2.Sept2000:
*Corresponding author. Tel.: 215-4669; fax: 00358-2-215-4806.
E-mail address:ralf.ostermark@abo."(R.OGstermark)
A nonlinear mixed integer multiperiod
"rm model
Ralf O
G
stermark
!
,
*
, Hans Skrifvars
"
, Tapio Westerlund
"
!Department of Business Administration, Asbo Akademi University, Henriksgatan 7, 20500 Turku, Finland
"Department of Chemical Engineering, Asbo Akademi University, Process Design Laboratory, Biskopsgatan 8, 20500 Turku, Finland Received 11 November 1996; accepted 27 January 2000
Abstract
We formulate and test an advanced multiperiod model for strategic"rm planning. This has not been previously considered as a mixed integer nonlinear problem (MINLP). Our approach is to show the di!erences between a linear and a nonlinear mixed integer approach. The key property of our model is the simplicity and e$ciency of generating optimal
"rm strategies, a cornerstone for managerial acceptance. Our purpose is to maximize the discounted value of net income and return on investment (ROI). Our model has been tested on some major Finnish"rms and it seems to give reliable results. With the data of our sample"rm for this paper, optimal ROI and optimal net income presuppose di!erent strategies. When optimizing ROI the model balances between cash and"xed assets, while optimizing net income results in an intensive investment program. Even if our sample "rm is but one case, the results are indicative of some fundamental principles governing managerial decision making. ( 2000 Elsevier Science B.V. All rights reserved.
Keywords: Strategic planning; Firm models; Linear programming; Optimization
1. Introduction
Mathematical experimenting is valuable for
planning
"
rm processes. Projections can be made
on future
"
nancial performance using a
"
nancial
analysis framework. Linear programming (LP) is
well known and widely used in business. Even
though, in many cases it is insu
$
cient to fully
capture the problem. Instead, many managerial
decision problems are of a mixed-integer character,
possibly containing various non-linearities. The
incentive for this study is to extend the framework
of So
K
derlund and O
G
stermark [1] to mixed integer
problems. We will subsequently maximize the
dis-counted net income (linear) and ROI (nonlinear) in
four cases as shown in Table 1.
Most
"
rm planning models use simulation to
project the consequences of alternative strategies
under a range of assumptions about the future.
However, these models do not provide the optimal,
i.e. the best, strategy, but only the consequences of
a strategy speci
"
ed by the user. The simulation
models primarily produce future accounting
state-ments, but there is no
"
nance theory to support
them. In this paper we are concerned with
optimiz-ing some main economic variables of the
"
rm over
a multi-year planning horizon, corresponding to
critical
"
nancing, operating and investment
deci-sions. The key property of our model is the
simpli-city in generating optimal
"
rm strategies.
The key contribution of our study is in the
derivation of a managerially convenient planning
0925-5273/00/$ - see front matter (2000 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 5 2 7 3 ( 0 0 ) 0 0 0 1 9 - 0
(2)
Table 1
The objective functions in the di!erent test settings Solution space
Decision variables Linear Nonlinear Continuous Net income (LP) ROI (NLP) Discrete Net income (MILP) ROI (MINLP)
system, requiring only a few input parameters for
producing full-
#
edged multiperiod
"
rm strategies.
The imprecision inherent in the parameters may be
recognized in Monte-Carlo-type scenarios (cf. e.g.
[2]). A further justi
"
cation for parsimony is the
speed of generating alternative optimal strategies.
Applications of linear programming models in
"
rm
planning are well documented. For example,
banking models have been widely developed within
operational research (cf. e.g. [3
}
7]). Reid and
Bradford [8] produced a farm
"
rm model of
ma-chinery investment decisions. The features of this
model have in
#
uenced various details of our design.
2. The MINLP-algorithm
The MINLP problem used in the method may be
formulated as follows:
min
x
,
y
Mc
T
x
x
#
c
T
y
yN
,
Mx
,
yN
3
N
,
(2.1)
where
N
"
Mx
,
yD
u
(
x
,
y
)
)
0N
.
(2.2)
c
x
and
c
y
are vectors with constants,
x
is a vector
with continuous variables,
y
is a vector with integer
variables and
u
(
x
,
y
) is a vector with nonlinear
dif-ferentiable functions, all de
"
ned on a set
¸"
X
X
>
,
(2.3)
where
X
is an
n
-dimensional compact polyhedral
convex set,
X
"
M
x
D
Ax
)
a
,
x
3
Rn
N
(2.4)
and
>
is a
"
nite discrete set de
"
ned by
>
"
M
y
D
By
)
b
,
y
3
Z
m
N
.
(2.5)
In the case of a nonlinear objective function
f
(
x
,
y
),
the objective function can be written as a nonlinear
constraint
f
(
x
,
y
)
!
k
)
0
(2.6)
and the corresponding variable
k
is minimized.
A sequence of points
M
(
x
k
,
y
k
),
k
"
0, 1,
2
,
K
N
generated by the ECP and
a
-ECP methods [9,10],
converging to the optimal solution of the problem
in Eq. (2.1) is given by
min
xk
,
ykMc
T
x
x
k
#
c
T
y
y
k
N
,
M
x
k
,
y
k
N
3
X
k
,
k
"
0, 1,
2
,
K
,
(2.7)
where
X
k
is de
"
ned by
X
K
"¸
WMx
,
yD
l
k
(
x
,
y
)
)
0,
k
"
0, 1,
2
,
K
!
1
N
,
(2.8)
where
l
k
(
x
,
y
)
"
f
k
(
x
k
,
y
k
)
#
a
k
)
AA
L
f
k
L
x
B
xk,
yk(
x
!
x
k
)
#
A
L
f
k
L
y
B
xk,
yk(
y
!
y
k
)
B
(2.9)
and
f
k
(
x
,
y
) is the function
g
i
(
x
,
y
) corresponding to
max
i
M
g
i
(
x
k
,
y
k
)
N
.
From the de
"
nition of the
X
k
set it follows from
Westerlund et al. [10] that the optimal solution
M
x
H,
y
H
N
of (2.1) is a subset of
X
k
for convex and
quasi-convex problems, and
X
K
L
X
K
~1
L2L
X
k
L2L
X
0
.
(2.10)
From Eq. (2.10) it also follows that the solutions
Z
k
"
min
Mc
T
x
x
k
#
c
T
y
y
k
N
form a monotonically
in-creasing sequence,
Z
K
*
Z
K
~1
*2*
Z
k
*2*
Z
0
.
(2.11)
The
convergence
of
the
sequence,
M
Z
k
,
k
"
0, 1,
2
,
K
N
, to the optimal solution in
N
is
shown for convex problems in [9], and for
quasi-convex problems in [10].
3. Model speci
5
cation
Our
"
rm model encompasses a planning horizon
(3)
Table 2
The objective functions, restrictions, decision variables and parameters of the"rm model Discounted
objectives
Restrictions Variables Parameters
O1 Net income R1 Sales}upper bound is a function of production capacity
#inventory
SALES
t Sales mindep Minimal depreciation R2 Amortization}equals
a proportion of long-term debt
PROD
t Production tax Taxes R3 New issues}upper bound is
a proportion of stockholders' equity
NEWDEBT
t New debt cost Operating costs/Turnover R4 Dividends}upper bound is
unrestricted shareholders equity AMO
t Amortization sr Sales receivable/Turnover R5 Dividends}lower bound is a
fraction of capital stock
INV
t Investments mxiss Maximal new issue/Stockholdersequity R6 Depreciation}lower bound is
a proportion of"xed assets
NEWISS
t New issue minequ Stockholders'equity/Liabilities R7 Equity/Debt}lower bound DIVID
t Dividends r Interest rate on long-term debt R8 Nonnegativity of cash DEP
t Depreciation cl Current liabilities/Operating costs R9 Nonnegativity of debt Deviation variables o" Other"nancial items/Other
"nancial assets R10 Nonnegativity of"xed assets AMODIFF in R2 d Discount factor
EQUITYDIFF in R3 X Machine cost MAXDIVDIFF in R4 p
t Unit sales price at timet MINDIVDIFF in R5 Factor Production capacity factor DEDIFF in R7 rep Amortization/Long-term debt
mindiv Minimal dividend/Capital stock
planning, where we are dealing with aggregate
deci-sions, a planning perspective of
"
ve years or longer
is usually desirable [[11], p. 713]. The set of
vari-ables is limited to those necessary for de
"
ning an
enterprise in economic terms. In order to represent
the
"
nancial variables accounting logic is used. The
"
nancing choice for an investment is an essential
decision. Should the management borrow, issue
new equity or use internal funds? The use of
retained earnings a
!
ects the capability of paying
dividends. The level of investments is a
!
ected by
the cost of capital. The level of sales and
deprecia-tion
}
both connected to investments
}
in
#
uence
net income. Like the interest expenses, the
deprecia-tions operate as a tax shield. The amortization rate
is determined schematically as a fraction of
accountable debt (see Table 2).
The key
"
nancial decision variables are
invest-ments, new loans, new issues, loan amortization,
dividend payments, depreciation, sales volume and
production volume. Since sales and production are
unsynchronized, inventory accumulation is
pos-sible. Inventory valuation is a crucial issue
involv-ing tax problems and matchinvolv-ing of income and
expenses within accounting periods. Furthermore,
a set of deviation variables guaranteeing solvability
is speci
"
ed below. The structure of the
"
nancial
statements is described in Tables 3 and 4.
The
"
nancial constraints include some
funda-mental requirements such as nonnegativity of
assets and liabilities and some economic conditions
and aspiration levels. The restrictions are based
on accounting legislation (e.g. maximal dividend),
on accounting
}
technical logic (e.g. nonnegativity
of assets and liabilities), on requirements imposed
by the economic environment (e.g. debt
}
equity
ratio and other relations) and on restrictions on
the productive capacity. Some restrictions are
allowed to diverge if the solution is infeasible.
Mathematically, this is settled by introducing
(4)
Table 3
Balance sheet of the"rm model
Assets Symbols Equations Liabilities Symbols Equations
Fixed assets
FIXASS
t FIXASSt~1#INVt!DEPt Capital stock EQUITYt EQUITYt~1#NEWISSt Sales
receivable
SALESREC
t sr TURNOt
"SALES
t*SPRICEt
Other unrestricted equity
OUEQUITY
t OUEQUITYt~2
#NETINCOME t~1 !DIVID t Cash and bank deposits CASH
t CASHt~1#NETINCOMEt
!DEP
t#SALESRECt
!SALESREC
t~1#OTFINASSt
!OTFINASS
t~1!INVt
#NEWISS
t!DIVIDt
!CURRLIAB
t#CURRLIABt~1
!AMO t
Net income NETINCOME
t See statement of income in Table 4
Other "nancial assets
OTFINASS
t OTFINASSt~1
#OTFINASS t
Shareholders' equity
TOTEQUITY
t EQUITYt~1
#OUEQUITY t
#NETINCOME t Current liabilities CURRLIAB
t cl cost TURNOt Long-term debt DEBT
t DEBTt~1#NEWDEBTt Total liabilities LIAB
t CURRLIABt#DEBTt Table 4
Statement of income
Item Symbols/Equations Item Equations
#Sales SALES
t !Interest payments r(DEBTt~1#NEWDEBTt)
!Production costs PROD
t #Other"nancial income o"]OTFINASSt
!Operating costs cost !Taxes tax][TURNOt(1!cost)!DEP t
!r](DEBTt~1#NEWDEBT
ti)#o"]OTFINASSt]
!Depreciation DEP
t NETINCOMEt r[1!tax]][TURNOt(1!cost)!DEPt!
](DEBTt~1#NEWDEBT
ti)#o"]OTFINASSt]
sanctioned deviation variables in the equations.
The optimization problem can be formulated as
follows:
MAX
x
ty
"
f
(
x
t
)
s.t.
A
t
x
t
)
b
t
,
t
"
1,
2
,
h
,
x
t
'
c
t
3
Rn
,
b
t
3
Rm
,
A
t
3
Rm
C
n
,
(3.1)
where
h
is the planning horizon. In the test below
we use
h
"
5.
For each period,
x
t
consists of the following 13
variables:
x@
t
"
(sales volume, production volume, new
debt, amortization, investments, new issue,
divi-dends, depreciation, amortization di
!
erence, equity
di
!
erence, maximal dividend di
!
erence, minimal
dividend di
!
erence, depreciation di
!
erence)
t
.
The objective functions are discounted to present
value by a discount factor (see Table 2 for a
description of the variables and parameters). The
variables are subject to 12 constraints as explained
below. For a multiobjective generalization of the
problem formulation, see, for example, [12,13].
(5)
3.1. The objecti
v
e functions
NETINCOME:
+
t
i
/1
C
(1
!
tax)[(1
!
cost)TURNO
i
!
r
+
t
j
/1
[NEWDEBT
ij
!
AMO
ij
]
!
DEP
i
]
(1
#
d
)
i
D
!
PENALTY
*
+
t
i
/1
5
+
j
/1
Dev
ij
,
(3.2)
ROI:
+
t
i
/1
NETINCOME
i
(1/(1
#
d
)
i
)
+
ti
/1
(INV
i
!
DEP
i
)/(1
#
d
)
i
#
FIXASS
0
!
PENALTY
*
+
t
i
/1
5
+
j
/1
Dev
ij
.
(3.3)
PENALTY is a positive value su
$
ciently large to
make deviations undesirable. Dev
ij
refers to the
deviation variables in the constants below.
3.2. The restrictions
The restrictions of our model are presented in
Table 2.
3.2.1. R1
:
The capacity constraint
Sales is de
"
ned as a function of production
capa-city of the
"
rm. Capacity, again, is related to
machinery as part of total
"
xed assets. The sales
value of production volume is as follows:
SALES VALUE OF PRODUCTION(
t
)
"
Factor
]
FIXASS
t
P
t
/unit
X
/unit
,
(3.4)
where the sales price,
P
t
, can vary over time. The
symbol
X
stands for the machine cost per unit. To
illustrate: assume that we have acquired a machine
for 1000 money units. The estimated production
of the machine over its entire lifetime is 5.000 units
of a certain product. The machine cost per unit
is then,
X
"
1000/5000
"
0.2. Assuming that this
machine is the only
"
xed asset (Factor
"
1;
FIXASS
t
"
1000), the sales value of production is
(1000/0.2)
P
t
"
5000
P
t
. The production costs are
assumed to be constant. Only the sales price per
unit (
P
t
) varies and it is assumed to be known. The
capacity constraint is formulated as follows:
SALES
t
(
Factor
]
P
t
X
C
FIXASS
0
#
+
t
i
/1
[INV
i
!
DEP
i
]
D
.
(3.5)
3.2.2. R2
:
Loan repayment
The second restriction concerns the level of
re-payment. The amortization amount equals, as far
as possible, a fraction of long-term debt, i.e.
AMO
t
"
rep
]
C
DEBT
0
#
t
+
i
/1
NEWDEBT
i
!
t
~1
+
i
/1
AMO
i
D
#
AMODIFF
t
.
(3.6)
The
"
rm is obliged to follow the plan of repayment,
but if necessary
}
for example, due to risk for
insolvency/bankruptcy
}
an exception is allowed.
In practice, the loan repayment schedule is of
course more complicated. Each loan has its own
amortization plan and repayments are not a
con-stant fraction of total debt. However, in the long
term, when the
"
rm approaches its equilibrium
level of operations, the total repayments will
amount to a fairly constant fraction of debts
out-standing. Before reaching this level, the repayments
could be modelled more exactly, for example, by
allowing the repayment fraction to vary over the
planning horizon.
3.2.3. R3
:
Upper bound on new issues
The upper bound on new shares is based on
rational economic arguments. The issue costs, the
possible negative reaction of the stock market (see,
for example, [14]) and the demand for new equity
limit the level of the issue. We restrict the new issues
to a fraction of stockholders
'
equity at the base year
(6)
as follows:
NEWISS
t
)
mxiss
]
EQUITY
0
#
EQUITYDIFF
t
.
(3.7)
The deviation variable allows new issues to diverge
if necessary. One may argue that there is also
a lower bound due to the
"
xed costs associated
with new issues. From a strategic planning point of
view, such costs are considered negligible, however.
In practice, the decision to issue new equity or to
prefer new debt, is governed by the target
equity/debt ratio or some other objective related to
controlling the
"
rm. Recognizing the
"
xed costs of
new issues (through appropriate binary-valued
variables) would make the model unduly
complic-ated in comparison to the expected utility.
3.2.4. R4
,
R5
:
Upper
/
lower bounds on di
v
idends
Dividends are limited by upper and lower
bounds. The upper bound is de
"
ned as free
un-restricted equity, i.e. retained earnings:
DIVID
t
(
OUEQUITY
0
#
t
+
i
/0
NETINCOME
i
!
t
~1
+
i
/1
DIVID
i
#
MAXDIVDIFF
t
. (3.8)
This is in order to protect creditors from
excess-ive dividend payments. A minimum level of
pay-ments is motivated by the shareholders
'
demand for
a stable dividend. A cutback of the dividend rate
would probably have a negative impact on the
market value of the
"
rm [15]. Especially minor
shareholders are guaranteed a fair dividend
through the lower bound:
DIVID
t
'
mindiv
C
EQUITY
0
#
t
+
i
/1
NEWISS
i
D
!
MINDIVDIFF
t
.
(3.9)
The deviation variable MAXDIVDIFF was
in-cluded to guarantee feasibility in cases where
unre-stricted equity is negative. In practice, the dividend
policy is much more complex than can be captured
by an upper and lower limit on dividends. Yet,
there is a signi
"
cant managerial interest in knowing
the
leeway
for dividend payment provided by the
optimal strategic scenario.
3.2.5. R6
:
Depreciations
The minimal depreciation level is governed by
economic and physical considerations. The
math-ematical expression is
DEP
t
'
mindep
C
FIXASS
0
#
t
+
i
/1
INV
i
!
t
~1
+
i
/1
DEP
i
D
.
(3.10)
3.2.6. R7
:
The equity
/
debt relation
This constraint controls the capital structure of
the
"
rm. According to Modigliani and Miller
'
s [16]
classical work on the theory of capital structure, the
mixture of
"
nancing investments does not a
!
ect the
value of the
"
rm in a world without taxes. When
taxes [17] and cost of bankruptcy [18] are
intro-duced, a trade-o
!
between these will lead to an
optimal capital structure (see also [19,20]). This
reasoning is partly supported by recent empirical
evidence, even though counterevidence does exist.
Firms with safe, tangible assets and plenty of
taxable income have higher debt-to-equity ratios
than an unpro
"
table and risky business with
intan-gible assets [21,22]). On the other hand, the
peck-ing order theory [23] explains why some pro
"
table
"
rms borrow less, as they do not need outside
money. Kjellman and Hanse
H
n [24] found that most
Finnish
"
nancial managers seek to maintain a
con-stant debt-to-equity ratio. A target debt ratio
is obviously a part of the
"
rms
' "
nancing policy.
In our model, the ratio is de
"
ned as the relation
between shareholders
'
equity and total liabilities:
EQUITY
0
#
t
+
i
/1
[NEWISS
i
#
NETINCOME
i
!
DIVID
i
]
#
DEDIFF
t
'
minequ
C
DEBT
0
#
CURRLIAB
t
#
+
t
i
/1
[NEWDEBT
i
!
AMO
i
]
D
.
(3.11)
Equity is made up of new issues
#
retained
earn-ings
!
dividends.
The above seven restrictions describe the relation
between the
"
rm and its environment. The next set
de
"
nes the necessary nonnegativity relations for the
(7)
Table 5
Parameters based on historical development
Symbols Values Symbols Values Symbols Values
mindep 0.06 o" 0.67 FIXASS
4 1078.00
tax 0.25 t 5 SALESREC
4 95.00
cost 1.00 X 1.00 CASH
4 12.00
sr 0.15 t"5 t"6 t"7 t"8 t"9 OTFINASS
4 30.00
d 0.15 P
t 1.05 1.10 1.30 1.20 1.25 EQUITY4 630.00
mxiss 0.02 Factor 0.80 NETINCOME
4 85.00
minequ 1.00 rep 0.08 OUEQUITY
4 0.00
r 0.08 mindiv 0.01 DEBT
4 600.00
cl 0.15 CURRLIAB
4 0.00
3.2.7. R8
:
Nonnegati
v
ity of cash
The cash
#
ow is de
"
ned as
C
CASH
0
#
[SALESREC
t
~1
!
SALESREC
t
]
#
[OTFINASS
t
~1
!
OTFINASS
t
]
!
[CURRLIAB
t
~1
!
CURRLIAB
t
]
#
+
ti
/1
[NETINCOME
i
#
DEP
i
!
INV
i
#
NEWDEBT
i
#
NEWISS
i
!
DIVID
i
]
D
*
0.
(3.12)
3.2.8. R9
:
Nonnegati
v
ity of debt
Nonnegativity of long-term debt is de
"
ned as
DEBT
t
"
DEBT
0
#
+
t
i
/1
[NEWDEBT
i
!
AMO
i
]
'
0.
(3.13)
Normally, debt will be positive while the
amortiza-tion rate is less than unity.
3.2.9. R10
:
Nonnegati
v
ity of
x
xed assets
The
"
nal constraint concerns nonnegativity of
"
xed assets:
FIXASS
t
"
FIXASS
0
#
t
+
i
/1
[INV
i
!
DEP
i
]
'
0.
(3.14)
If selling of
"
xed assets is allowed, this must be
included in the restriction. Otherwise it is
redund-ant, since depreciation is always a nonnegative
frac-tion of total assets.
4. A numerical experiment
Our model has been tested on data from some
Finnish listed companies. A hypothetical
"
rm is
studied below over a
"
ve year planning horizon.
The upper bound on sales volume is determined by
new investments, since the size of
"
xed assets
deter-mines the capacity of production. New investments,
again, are restricted by the
"
nancial structure of the
"
rm. The
"
nancing alternatives are internal or
external sources of funds, i.e. retained earnings, new
equity and new debt. Depreciation a
!
ects the
econ-omy of the
"
rm in three di
!
erent ways. Firstly, it
reduces the maximal allowed dividends. Secondly,
it decreases the value of
"
xed assets which a
!
ects
production capacity and reduces net income.
Thirdly, it provides a tax shield a
!
ecting cash
#
ows.
If new issues and internal funds do not, however,
su
$
ce to
"
nance new investment, the
"
rm is forced
to borrow. When maximizing net income the
capi-tal structure is particularly sensitive to the interest
rate and investments and new debts are negatively
correlated with the interest rate.
The base year (year 4)
"
nancial statements and
control parameters are presented in Table 5.
The parameters are speci
"
ed on the basis of
historical values (years 0
}
4) and a subjective
judge-ment of future developjudge-ment. The historical
state-ments along with the optimal projected statestate-ments
and the optimal programs are given in Tables 6
}
9.
(8)
Table 6
Maximizing net income in the linear continuous case (LP)
Example Optimization
Value of objective function 772.15
Decision variables 5 6 7 8 9
1: Sales 733.25 787.92 1307.51 1290.09 1490.94
2: Production 810.66 776.43 1141.61 1290.09 1490.94
3: New debt 0.00 0.00 145.44 165.50 172.04
4: Amortization 48.00 44.16 52.26 61.32 70.18
5: Investments 0.00 19.16 547.56 288.54 370.01
6: New issues 14.30 14.30 14.30 14.30 14.30
7: Dividends 6.44 6.59 6.73 6.87 7.02
8: Depreciation 64.68 61.95 91.09 102.93 118.96
9: Dividend deviation (Divdi!) 0.00 0.00 0.00 0.00 0.00
10: Equity deviation (EKdi!) 0.00 0.00 0.00 0.00 0.00
11: Debt-Equity deviation (D/E}di!) 0.00 0.00 0.00 0.00 0.00
12: Repayment deviation (REPdi!) 0.00 0.00 0.00 0.00 0.00
13: Max dividend deviation (MAXdivdf)
0.00 0.00 0.00 0.00 0.00
Historical period: 0 1 2 3 4
Inventory volume 17 8 8 8 100
Planning period: 5 6 7 8 9
Inventory volume 177.40 165.91 0.00 0.00 0.00
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Assets
Fixed assets 859.00 914.00 1016.00 1070.00 1078.00 1013.32 970.53 1427.01 1612.62 1863.67 Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Inventory 10.00 5.00 5.00 5.00 100.00 177.40 165.91 0.00 0.00 0.00 Sales receivables 150.00 180.00 170.00 190.00 95.00 115.49 130.01 254.97 232.22 279.55
Cash 10.00 15.00 15.00 15.00 12.00 0.00 0.00 0.00 0.00 0.00
Other"nancial assets 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 Financial assets 190.00 225.00 215.00 235.00 137.00 145.49 160.01 284.97 262.22 309.55
Assets 1059.00 1144.00 1236.00 1310.00 1315.00 1336.21 1296.45 1711.97 1874.83 2173.22 Shares equity and liabilities
Capital stock 500.00 500.00 550.00 550.00 630.00 644.30 658.60 672.90 687.20 701.50 Other restricted equity 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Other unrestricted equity 0.00 0.00 0.00 0.00 0.00 78.56 17.84 !6.73 176.21 243.20 Net income for the year 109.00 84.00 71.00 85.00 85.00 !54.13 !17.84 189.82 74.00 141.91 Shareholders'equity 609.00 584.00 621.00 635.00 715.00 668.72 658.60 855.99 937.42 1086.61 Accumulated depreciation di!erence 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Reserves 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Current liabilities 50.00 60.00 65.00 75.00 0.00 115.49 130.01 254.97 232.22 279.55 Long-term debt 400.00 500.00 550.00 600.00 600.00 552.00 507.84 601.02 705.20 807.06 Liabilities 450.00 560.00 615.00 675.00 600.00 667.49 637.85 855.99 937.42 1086.61
(9)
Table 6 (continued)
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Statement of income
Turnover 500.00 510.00 510.00 550.00 600.00 769.91 866.71 1699.77 1548.11 1863.67 Operating costs 300.00 321.00 324.00 340.00 380.00 733.25 787.92 1307.51 1290.09 1490.94 Operating income 200.00 189.00 186.00 210.00 220.00 36.66 78.79 392.25 258.02 372.73 Depreciation 50.00 55.00 60.00 65.00 70.00 64.68 61.95 91.09 102.93 118.96 Operating income after depreciation 150.00 134.00 126.00 145.00 150.00 !28.02 16.84 301.17 155.09 253.78 Interest expenses !40.00 !50.00 !55.00 !58.00 !62.00 !44.16 !40.63 !48.08 !56.42 !64.56 Other"nancial income 20.00 20.00 20.00 20.00 20.00 0.00 0.00 0.00 0.00 0.00
Extraordinary income and expenses 0.00 0.00 0.00 0.00 0.00
Allocations 0.00 0.00 0.00 0.00 0.00
Taxes 21.00 20.00 20.00 22.00 23.00 !18.04 !5.95 63.27 24.67 47.30
Net income 109.00 84.00 71.00 85.00 85.00 !54.13 !17.84 189.82 74.00 141.91
Other information
Amortization 30 40 50 50 55 48.00 44.16 52.26 61.32 70.18
Investments 110 162 119 78 0.00 19.16 547.56 288.54 370.01
New issues 0 50 0 80 14.3 14.3 14.3 14.3 14.3
Dividends 109 84 71 85 0 6.443 6.586 6.729 6.872 7.015
Table 7
Maxmizing net income in the linear mixed-integer case (MLP)
Example Optimization
Value of objective function 772.60
Decision variables 5 6 7 8 9
1: Sales 734.00 782.00 1311.00 1291.00 1493.00
2: Production 811.00 774.00 1142.00 1291.00 1493.00
3: New debt 0.00 0.00 145.05 166.09 172.58
4: Amortization 48.00 44.16 52.23 61.34 70.24
5: Investments 0.46 15.51 551.12 289.26 371.62
6: New issues 14.30 14.30 14.30 14.30 14.30
7: Dividends 6.44 6.59 6.73 6.87 7.02
8: Depreciation 64.71 61.76 91.12 103.01 119.12
9: Dividend deviation (Divdi!) 0.00 0.00 0.00 0.00 0.00
10: Equity deviation (EKdi!) 0.00 0.00 0.00 0.00 0.00
11: Debt-Equity deviation (D/E}di!) 0.00 0.00 0.00 0.00 0.00
12: Repayment deviation (REPdi!) 0.00 0.00 0.00 0.00 0.00
13: Max dividend deviation (MAXdivdf)
0.00 0.00 0.00 0.00 0.00
Historical period: 0 1 2 3 4
Inventory volume 17 8 8 8 100
Planning period: 5 6 7 8 9
(10)
Table 7 (continued)
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Assets
Fixed assets 859.00 914.00 1016.00 1070.00 1078.00 1013.75 967.50 1427.50 1613.75 1866.25
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Inventory 10.00 5.00 5.00 5.00 100.00 177.00 169.00 0.00 0.00 0.00 Sales receivables 150.00 180.00 170.00 190.00 95.00 115.61 129.03 255.65 232.38 279.94 Cash 10.00 15.00 15.00 15.00 12.00 !0.02 !0.35 !0.37 !0.37 !1.11 Other"nancial assets 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 Financial assets 190.00 225.00 215.00 235.00 137.00 145.59 158.68 285.28 262.01 308.82
Assets 1059.00 1144.00 1236.00 1310.00 1315.00 1336.34 1295.18 1712.78 1875.76 2175.07 Shares equity and liabilities
Capital stock 500.00 500.00 550.00 550.00 630.00 644.30 658.60 672.90 687.20 701.50 Other restricted equity 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Other unrestricted equity 0.00 0.00 0.00 0.00 0.00 78.56 17.85 !7.02 176.21 243.76 Net income for the year 109.00 84.00 71.00 85.00 85.00 !54.13 !18.14 190.60 74.07 142.13 Shareholders'equity 609.00 584.00 621.00 635.00 715.00 668.73 658.31 856.48 937.98 1087.39 Accumulated depreciation
di!erence
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Reserves 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Current liabilities 50.00 60.00 65.00 75.00 0.00 115.61 129.03 255.65 232.38 279.94 Long-term debt 400.00 500.00 550.00 600.00 600.00 552.00 507.84 600.66 705.41 807.74 Liabilities 450.00 560.00 615.00 675.00 600.00 667.61 636.87 856.30 937.79 1087.68
Liabilities and shareholders+equity 1059.00 1144.00 1236.00 1310.00 1315.00 1336.34 1295.18 1712.78 1875.76 2175.07 Statement of income
Turnover 500.00 510.00 510.00 550.00 600.00 770.70 860.20 1704.30 1549.20 1866.25 Operating costs 300.00 321.00 324.00 340.00 380.00 734.00 782.00 1311.00 1291.00 1493.00 Operating income 200.00 189.00 186.00 210.00 220.00 36.70 78.20 393.30 258.20 373.25 Depreciation 50.00 55.00 60.00 65.00 70.00 64.71 61.76 91.12 103.01 119.12 Operating income after
depreciation
150.00 134.00 126.00 145.00 150.00 !28.01 16.44 302.18 155.19 254.13 Interest expenses !40.00 !50.00 !55.00 !58.00 !62.00 !44.16 !40.63 !48.05 !56.43 !64.62 Other"nancial income 20.00 20.00 20.00 20.00 20.00 0.00 0.00 0.00 0.00 0.00 Extraordinary income and
expenses
0.00 0.00 0.00 0.00 0.00
Allocations 0.00 0.00 0.00 0.00 0.00
Taxes 21.00 20.00 20.00 22.00 23.00 !18.04 !6.05 63.53 24.69 47.38
Net income 109.00 84.00 71.00 85.00 85.00 !54.13 !18.14 190.60 74.07 142.13
Other information
Amortization 30 40 50 50 55 48.00 44.16 52.23 61.34 70.24
Investments 110 162 119 78 0.46 15.51 551.12 289.26 371.62
New issues 0 50 0 80 14.3 14.3 14.3 14.3 14.3
(11)
Table 8
Maxmizing ROI in the nonlinear continuous case (NLP)
Example Optimization
Value of objective function 0.65
Decision variables 5 6 7 8 9
1: Sales 731.68 784.60 1311.52 833.73 1249.45
2: Production 810.00 774.80 1142.35 1073.81 1009.38
3: New debt 1.46 0.00 144.29 193.62 118.01
4: Amortization 48.12 44.27 52.27 63.58 67.93
5: Investments 0.00 17.00 550.58 0.00 0.00
6: New issues 14.30 14.30 14.30 14.30 14.30
7: Dividends 6.44 6.59 6.73 6.87 7.02
8: Depreciation 64.68 61.82 91.14 85.68 80.54
9: Dividend deviation (Divdi!) 0.00 0.00 0.00 0.00 0.00
10: Equity deviation (EKdi!) 0.00 0.00 0.00 0.00 0.00
11: Debt-Equity deviation (D/E}di!) 0.00 0.00 0.00 0.00 0.00
12: Repayment deviation (REPdi!) 0.00 0.00 0.00 0.00 0.00
13: Max dividend deviation (MAXdivdf)
0.00 0.00 0.00 0.00 0.00
Historical period: 0 1 2 3 4
Inventory volume 17 8 8 8 100
Planning period: 5 6 7 8 9
Inventory volume 178.98 169.18 0.01 240.09 0.02
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Assets
Fixed assets 859.00 914.00 1016.00 1070.00 1078.00 1013.32 968.50 1427.93 1342.26 1261.72
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Inventory 10.00 5.00 5.00 5.00 100.00 178.98 169.18 0.01 240.09 0.02 Sales receivables 150.00 180.00 170.00 190.00 95.00 115.24 129.46 255.75 150.07 234.27 Cash 10.00 15.00 15.00 15.00 12.00 !0.37 !0.37 !0.37 !0.38 504.59 Other"nancial assets 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 Financial assets 190.00 225.00 215.00 235.00 137.00 144.87 159.09 285.37 179.69 768.86
Assets 1059.00 1144.00 1236.00 1310.00 1315.00 1337.17 1296.77 1713.31 1762.04 2030.60 Shares equity and liabilities
Capital stock 500.00 500.00 550.00 550.00 630.00 644.30 658.60 672.90 687.20 701.50 Other restricted equity 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Other unrestricted equity 0.00 0.00 0.00 0.00 0.00 78.56 17.70 !7.09 176.70 186.62
Net income for the year 109.00 84.00 71.00 85.00 85.00 !54.27 !18.06 190.67 16.93 127.00 Shareholders'equity 609.00 584.00 621.00 635.00 715.00 668.58 658.23 856.47 880.84 1015.12 Accumulated depreciation
di!erence
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Reserves 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Current liabilities 50.00 60.00 65.00 75.00 0.00 115.24 129.46 255.75 150.07 234.27 Long-term debt 400.00 500.00 550.00 600.00 600.00 553.34 509.08 601.09 731.13 781.21 Liabilities 450.00 560.00 615.00 675.00 600.00 668.58 638.54 856.84 881.20 1015.48
(12)
Table 8 (continued)
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Statement of income
Turnover 500.00 510.00 510.00 550.00 600.00 768.27 863.06 1704.98 1000.47 1561.81 Operating costs 300.00 321.00 324.00 340.00 380.00 731.68 784.60 1311.52 833.73 1249.45 Operating income 200.00 189.00 186.00 210.00 220.00 36.58 78.46 393.46 166.75 312.36 Depreciation 50.00 55.00 60.00 65.00 70.00 64.68 61.82 91.14 85.68 80.54 Operating income after
depreciation
150.00 134.00 126.00 145.00 150.00 !28.10 16.64 302.31 81.07 231.83 Interest expenses !40.00 !50.00 !55.00 !58.00 !62.00 !44.27 !40.73 !48.09 !58.49 !62.50 Other"nancial income 20.00 20.00 20.00 20.00 20.00 0.00 0.00 0.00 0.00 0.00 Extraordinary income and
expenses
0.00 0.00 0.00 0.00 0.00
Allocations 0.00 0.00 0.00 0.00 0.00
Taxes 21.00 20.00 20.00 22.00 23.00 !18.09 !6.02 63.56 5.64 42.33
Net income 109.00 84.00 71.00 85.00 85.00 !54.27 !18.06 190.67 16.93 127.00
Other information
Amortization 30 40 50 50 55 48.12 44.27 52.27 63.58 67.93
Investments 110 162 119 78 0.00 17.00 550.58 0.00 0.00
New Issues 0 50 0 80 14.3 14.3 14.3 14.3 14.3
Dividends 109 84 71 85 0 6.443 6.586 6.729 6.872 7.015
Table 9
Maxmizing ROI in the nonlinear mixed-integer case (MINLP)
Example Optimization
Value of objective function 0.65
Decision variables 5 6 7 8 9
1: Sales 731.00 785.00 1310.00 834.00 1249.00
2: Production 810.00 774.00 1142.00 1073.00 1010.00
3: New debt 1.55 0.78 143.47 193.32 118.02
4: Amortization 48.12 44.34 52.27 63.55 67.91
5: Investments 0.00 15.94 551.12 0.00 1.24
6: New issues 14.30 14.30 14.30 14.30 14.30
7: Dividends 6.44 6.59 6.73 6.87 7.02
8: Depreciation 64.68 61.76 91.12 85.65 80.59
9: Dividend deviation (Divdi!) 0.00 0.00 0.00 0.00 0.00
10: Equity deviation (EKdi!) 0.00 0.00 0.00 0.00 0.00
11: Debt-Equity deviation (D/E}di!) 0.00 0.00 0.00 0.00 0.00
12: Repayment deviation (REPdi!) 0.00 0.00 0.00 0.00 0.00
13: Max dividend deviation (MAXdivdf)
0.00 0.00 0.00 0.00 0.00
Historical period: 0 1 2 3 4
Inventory volume 17 8 8 8 100
Planning period: 5 6 7 8 9
(13)
Table 9 (continued)
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Assets
Fixed assets 859.00 914.00 1016.00 1070.00 1078.00 1013.32 967.50 1427.50 1341.85 1262.50
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Inventory 10.00 5.00 5.00 5.00 100.00 179.00 168.00 0.00 239.00 0.00 Sales receivables 150.00 180.00 170.00 190.00 95.00 115.13 129.53 255.45 150.12 234.19
Cash 10.00 15.00 15.00 15.00 12.00 !0.34 2.60 !0.28 0.58 503.22
Other"nancial assets 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 Financial assets 190.00 225.00 215.00 235.00 137.00 144.79 162.13 285.17 180.70 767.41
Assets 1059.00 1144.00 1236.00 1310.00 1315.00 1337.11 1297.63 1712.67 1761.55 2029.91 Shares equity and liabilities
Capital stock 500.00 500.00 550.00 550.00 630.00 644.30 658.60 672.90 687.20 701.50 Other restricted equity 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Other unrestricted equity 0.00 0.00 0.00 0.00 0.00 78.56 17.67 !7.09 176.38 186.38 Net income for the year 109.00 84.00 71.00 85.00 85.00 !54.30 !18.03 190.35 17.01 126.89 Shareholders'equity 609.00 584.00 621.00 635.00 715.00 668.55 658.23 856.15 880.59 1014.77 Accumulated depreciation
di!erence
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Reserves 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Current liabilities 50.00 60.00 65.00 75.00 0.00 115.13 129.53 255.45 150.12 234.19 Long-term debt 400.00 500.00 550.00 600.00 600.00 553.42 509.87 601.07 730.84 780.95 Liabilities 450.00 560.00 615.00 675.00 600.00 668.55 639.40 856.52 880.96 1015.14
Liabilities and shareholders+equity 1059.00 1144.00 1236.00 1310.00 1315.00 1337.11 1297.63 1712.67 1761.55 2029.91 Statement of income
Turnover 500.00 510.00 510.00 550.00 600.00 767.55 863.50 1703.00 1000.80 1561.25 Operating costs 300.00 321.00 324.00 340.00 380.00 731.00 785.00 1310.00 834.00 1249.00 Operating income 200.00 189.00 186.00 210.00 220.00 36.55 78.50 393.00 166.80 312.25 Depreciation 50.00 55.00 60.00 65.00 70.00 64.68 61.76 91.12 85.65 80.59 Operating income after
depreciation
150.00 134.00 126.00 145.00 150.00 !28.13 16.74 301.88 81.15 231.66 Interest expenses !40.00 !50.00 !55.00 !58.00 !62.00 !44.27 !40.79 !48.09 !58.47 !62.48 Other"nancial income 20.00 20.00 20.00 20.00 20.00 0.00 0.00 0.00 0.00 0.00 Extraordinary income and
expenses
0.00 0.00 0.00 0.00 0.00
Allocations 0.00 0.00 0.00 0.00 0.00
Taxes 21.00 20.00 20.00 22.00 23.00 !18.10 !6.01 63.45 5.67 42.30
Net income 109.00 84.00 71.00 85.00 85.00 !54.30 !18.03 190.35 17.01 126.89
Other information
Amortization 30 40 50 50 55 48.12 44.34 52.27 63.55 67.91
Investments 110 162 119 78 0.00 15.94 551.12 0.00 1.24
New issues 0 50 0 80 14.3 14.3 14.3 14.3 14.3
(14)
Fig. 2. Maximizing net income (NI) in the mixed-integer linear case (cf. Table 7). Fig. 1. Maximizing net income (NI) in the continous linear case (cf. Table 6).
The optimal trajectories for the critical success
fac-tors are shown in Figs. 1
}
8 .
The results show that the discrete solutions are
slightly better than the continuous ones for the last
planning period, due to rounding errors in the cash
position. Some rounding errors are also observed
in the cash position for the nonlinear continuous
solution.
5. Conclusion
In the present study a fundamental assumption
has been that the sales versus production decisions
of the
"
rm do not a
!
ect market demand, in
consist-ence with perfect competition. Thus, the sales of the
"
rm are limited only by the internal conditions of
the
"
rm, in particular, productive capacity and
inventory volume. To allow for market
imperfec-tions, we may constrain the sales of the
"
rm
through a demand constraint. A particular demand
relation is given by the well-known constant
elas-ticity of demand (CED) function
q
D
(
t
)
"
A
(
t
)
p
(
t
)
e
D,
(5.1)
where
A
(
t
) is a time-varying parameter estimated
from historic data of the
"
rm,
e
D
"
L
q
D
(
t
)
L
p
(
t
)
p
(
t
)
(15)
Fig. 3. Maximizing ROI in the nonlinear continous case (cf. Table 8).
Fig. 4. Maximizing ROI in the nonlinear mixed-integer case (cf. Table 9).
(16)
Fig. 6. ROI trajectory when maximizing net income (NI) in the mixed-integer linear case (cf. Table 7).
Fig. 7. NI trajectory when maximizing ROI in the continous linear case (cf. Table 8).
(17)
is the (constant) price elasticity of demand. Its
time-varying counterpart with nonconstant price
elastic-ity
e
D(
t
)
may also be used. Tenhunen [25] tested the
CED-function on Rautaruukki, a state-owned
Finnish steel manufacturing
"
rm, with yearly data
between 1990 and 1993. The estimated steel
quant-ities corresponded well with the realized
"
gures.
There are many possibilities to further re
"
ne and
develop our model. The riskiness of the business
environment can be recognized, e.g. by
Monte-Carlo simulations in the spirit of Kasanen et al. [13].
The impact of o
!
-balance sheet factors, such as
derivatives, is also relevant. Bessler and Booth [7]
have developed a bank model including derivative
securities. Another direction would be to concentrate
on techno-economic
"
rm planning, i.e. on
simulta-neous modelling of strategic decisions of the
"
rm and
calibration of its technical processes. Finally, our
"
rm-model could be extended to (multinational)
concerns, an important and worthwhile exercise.
Our results show that simple rounding of the
continuous solutions does not guarantee an optimal
mixed-integer solution in strategic
"
rm planning.
The problem with rounding errors in the cash
posi-tion deserves further attenposi-tion in future research.
Acknowledgements
Financial support from the Academy of Finland
is gratefully acknowledged.
References
[1] K. SoKderlund, R. OGstermark, A multiperiod"rm model for strategic decision support, Working paper, Asbo Akademi University, 1995.
[2] E. Kasanen, R. OGstermark, The managerial viewpoint in interactive programming with multiple objectives, Kyber-netes 16 (1987) 235}240.
[3] G.G. Booth, P.E. Kovesos, A programming model for bank hedging decisions, Journal of Financial Research 9 (1986) 271}279.
[4] H. Meyer zu Zelhausen, Commercial bank balance sheet optimization. A decision model approach, Journal of Banking and Finance 10 (1986) 119}142.
[5] A. Korhonen, A dynamic bank portfolio planning model with multiple scenarios, multiple goals and changing pri-orities, European Journal of Operational Research 30 (1987) 13}23.
[6] G.G. Booth, W. Bessler, W.G. Foote, Managing interest rate risk in banking institutions, European Journal of Operational Research 41 (1989) 302}313.
[7] W. Bessler, G.G. Booth, An interest rate risk management model for commercial banks, European Journal of Opera-tional Research 74 (1994) 243}256.
[8] D.W. Reid, G.L. Bradford, A"rm farm model of machin-ery investment decisions, American Journal of Agricul-tural Economics 69 (1987) 66}87.
[9] T. Westerlund, F. Pettersson, An extended cutting plane method for solving convex MINLP problems, Computers and Chemical Engineering 19 (Suppl.) (1995) S131}136. [10] T. Westerlund, H. Skrifvars, I. Harjunkoski, An extended
cutting plane method for a class of non-convex MINLP problems, Computers and Chemical Engineering 22 (1998) 357}365.
[11] R.A. Brealey, S.C. Myers, Principles of Corporate Finance, 4th Edition, McGraw-Hill, New York, 1991.
[12] R. OGstermark, Solving a linear multiperiod portfolio prob-lem by interior point methodology, Computer Science in Economics and Management 5 (1991) 283}302. [13] E. Kasanen, M. Zeleny, R. OGstermark, Gestalt system
of holistic graphics: New management support view of MCDM, in: A.G. Lockett, G. Islei (Eds.), Improving Deci-sion Making in Organizations, Lecture Notes in Econ-omics and Mathematical Systems, Springer, Berlin, 1989. [14] P. Asquith, D.W. Mullins, Equity issues and o!ering
dilu-tion, Journal of Financial Economics 15 (1986) 61}90. [15] P. Healey, K. Palepu, Earnings information conveyed by
dividend initiations and omissions, Journal of Financial Economics 21 (1988) 149}175.
[16] F. Modigliani, M.H. Miller, The cost of capital,
corpora-tion "nance and the theory of investment, American
Economic Review 48 (1958) 261}297.
[17] F. Modigliani, M.H. Miller, Corporate income taxes and cost of capital: A correction, American Economic Review 53 (1963) 433}443.
[18] J.B. Warner, Bankruptcy costs: Some evidence, Journal of Finance 32 (1977) 337}348.
[19] M.J. Gordon, Towards a theory of"nancial distress, Jour-nal of Finance 26 (1971) 337}348.
[20] M. Jensen, W. Meckling, Theory of the"rm: Managerial behaviour, agency costs, and ownership structure, Journal of Financial Economics 3 (1976) 305}360.
[21] J.B. Warner, Bankruptcy, absolute priority, and the pric-ing of risky debt claims, Journal of Financial Economics 4 (1977) 239}276.
[22] E. Altman, A further empirical investigation of the bank-ruptcy cost question, Journal of Finance 39 (1984) 1067}1089. [23] S.C. Myers, The capital structure puzzle, Journal of
Finance 39 (1984) 575}592.
[24] A. Kjellman, S. HanseHn, Determinants of capital structure: Theory vs practice, The Scandinavian Journal of Manage-ment 2 (1993) 91}102.
[25] L. Tenhunen, Supply functions and pricing behaviour in the theory of the"rm, University of Tampere. School of Busi-ness Administration, 1994, series A2. ISBN 951-44-3599-0.
(1)
Table 8 (continued)
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Statement of income
Turnover 500.00 510.00 510.00 550.00 600.00 768.27 863.06 1704.98 1000.47 1561.81 Operating costs 300.00 321.00 324.00 340.00 380.00 731.68 784.60 1311.52 833.73 1249.45 Operating income 200.00 189.00 186.00 210.00 220.00 36.58 78.46 393.46 166.75 312.36
Depreciation 50.00 55.00 60.00 65.00 70.00 64.68 61.82 91.14 85.68 80.54
Operating income after depreciation
150.00 134.00 126.00 145.00 150.00 !28.10 16.64 302.31 81.07 231.83 Interest expenses !40.00 !50.00 !55.00 !58.00 !62.00 !44.27 !40.73 !48.09 !58.49 !62.50
Other"nancial income 20.00 20.00 20.00 20.00 20.00 0.00 0.00 0.00 0.00 0.00
Extraordinary income and expenses
0.00 0.00 0.00 0.00 0.00
Allocations 0.00 0.00 0.00 0.00 0.00
Taxes 21.00 20.00 20.00 22.00 23.00 !18.09 !6.02 63.56 5.64 42.33
Net income 109.00 84.00 71.00 85.00 85.00 !54.27 !18.06 190.67 16.93 127.00
Other information
Amortization 30 40 50 50 55 48.12 44.27 52.27 63.58 67.93
Investments 110 162 119 78 0.00 17.00 550.58 0.00 0.00
New Issues 0 50 0 80 14.3 14.3 14.3 14.3 14.3
Dividends 109 84 71 85 0 6.443 6.586 6.729 6.872 7.015
Table 9
Maxmizing ROI in the nonlinear mixed-integer case (MINLP)
Example Optimization
Value of objective function 0.65
Decision variables 5 6 7 8 9
1: Sales 731.00 785.00 1310.00 834.00 1249.00
2: Production 810.00 774.00 1142.00 1073.00 1010.00
3: New debt 1.55 0.78 143.47 193.32 118.02
4: Amortization 48.12 44.34 52.27 63.55 67.91
5: Investments 0.00 15.94 551.12 0.00 1.24
6: New issues 14.30 14.30 14.30 14.30 14.30
7: Dividends 6.44 6.59 6.73 6.87 7.02
8: Depreciation 64.68 61.76 91.12 85.65 80.59
9: Dividend deviation (Divdi!) 0.00 0.00 0.00 0.00 0.00
10: Equity deviation (EKdi!) 0.00 0.00 0.00 0.00 0.00
11: Debt-Equity deviation (D/E}di!) 0.00 0.00 0.00 0.00 0.00
12: Repayment deviation (REPdi!) 0.00 0.00 0.00 0.00 0.00
13: Max dividend deviation (MAXdivdf)
0.00 0.00 0.00 0.00 0.00
Historical period: 0 1 2 3 4
Inventory volume 17 8 8 8 100
Planning period: 5 6 7 8 9
(2)
Table 9 (continued)
Financial statements Historical accounts Forecasted accounts
0 1 2 3 4 5 6 7 8 9
Assets
Fixed assets 859.00 914.00 1016.00 1070.00 1078.00 1013.32 967.50 1427.50 1341.85 1262.50
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Inventory 10.00 5.00 5.00 5.00 100.00 179.00 168.00 0.00 239.00 0.00
Sales receivables 150.00 180.00 170.00 190.00 95.00 115.13 129.53 255.45 150.12 234.19
Cash 10.00 15.00 15.00 15.00 12.00 !0.34 2.60 !0.28 0.58 503.22
Other"nancial assets 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 30.00 Financial assets 190.00 225.00 215.00 235.00 137.00 144.79 162.13 285.17 180.70 767.41
Assets 1059.00 1144.00 1236.00 1310.00 1315.00 1337.11 1297.63 1712.67 1761.55 2029.91
Shares equity and liabilities
Capital stock 500.00 500.00 550.00 550.00 630.00 644.30 658.60 672.90 687.20 701.50
Other restricted equity 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Other unrestricted equity 0.00 0.00 0.00 0.00 0.00 78.56 17.67 !7.09 176.38 186.38 Net income for the year 109.00 84.00 71.00 85.00 85.00 !54.30 !18.03 190.35 17.01 126.89 Shareholders'equity 609.00 584.00 621.00 635.00 715.00 668.55 658.23 856.15 880.59 1014.77 Accumulated depreciation
di!erence
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Reserves 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Valuation items 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Current liabilities 50.00 60.00 65.00 75.00 0.00 115.13 129.53 255.45 150.12 234.19 Long-term debt 400.00 500.00 550.00 600.00 600.00 553.42 509.87 601.07 730.84 780.95 Liabilities 450.00 560.00 615.00 675.00 600.00 668.55 639.40 856.52 880.96 1015.14
Liabilities and shareholders+equity 1059.00 1144.00 1236.00 1310.00 1315.00 1337.11 1297.63 1712.67 1761.55 2029.91
Statement of income
Turnover 500.00 510.00 510.00 550.00 600.00 767.55 863.50 1703.00 1000.80 1561.25 Operating costs 300.00 321.00 324.00 340.00 380.00 731.00 785.00 1310.00 834.00 1249.00 Operating income 200.00 189.00 186.00 210.00 220.00 36.55 78.50 393.00 166.80 312.25
Depreciation 50.00 55.00 60.00 65.00 70.00 64.68 61.76 91.12 85.65 80.59
Operating income after depreciation
150.00 134.00 126.00 145.00 150.00 !28.13 16.74 301.88 81.15 231.66 Interest expenses !40.00 !50.00 !55.00 !58.00 !62.00 !44.27 !40.79 !48.09 !58.47 !62.48
Other"nancial income 20.00 20.00 20.00 20.00 20.00 0.00 0.00 0.00 0.00 0.00
Extraordinary income and expenses
0.00 0.00 0.00 0.00 0.00
Allocations 0.00 0.00 0.00 0.00 0.00
Taxes 21.00 20.00 20.00 22.00 23.00 !18.10 !6.01 63.45 5.67 42.30
Net income 109.00 84.00 71.00 85.00 85.00 !54.30 !18.03 190.35 17.01 126.89
Other information
Amortization 30 40 50 50 55 48.12 44.34 52.27 63.55 67.91
Investments 110 162 119 78 0.00 15.94 551.12 0.00 1.24
New issues 0 50 0 80 14.3 14.3 14.3 14.3 14.3
(3)
Fig. 2. Maximizing net income (NI) in the mixed-integer linear case (cf. Table 7). Fig. 1. Maximizing net income (NI) in the continous linear case (cf. Table 6).
The optimal trajectories for the critical success
fac-tors are shown in Figs. 1
}
8 .
The results show that the discrete solutions are
slightly better than the continuous ones for the last
planning period, due to rounding errors in the cash
position. Some rounding errors are also observed
in the cash position for the nonlinear continuous
solution.
5. Conclusion
In the present study a fundamental assumption
has been that the sales versus production decisions
of the
"
rm do not a
!
ect market demand, in
consist-ence with perfect competition. Thus, the sales of the
"
rm are limited only by the internal conditions of
the
"
rm, in particular, productive capacity and
inventory volume. To allow for market
imperfec-tions, we may constrain the sales of the
"
rm
through a demand constraint. A particular demand
relation is given by the well-known constant
elas-ticity of demand (CED) function
q
D
(t)
"
A(t)p(t)
e
D,
(5.1)
where
A(t) is a time-varying parameter estimated
from historic data of the
"
rm,
e
D
"
L
q
D
(t)
L
p(t)
p(t)
(4)
Fig. 3. Maximizing ROI in the nonlinear continous case (cf. Table 8).
Fig. 4. Maximizing ROI in the nonlinear mixed-integer case (cf. Table 9).
(5)
Fig. 6. ROI trajectory when maximizing net income (NI) in the mixed-integer linear case (cf. Table 7).
Fig. 7. NI trajectory when maximizing ROI in the continous linear case (cf. Table 8).
(6)
is the (constant) price elasticity of demand. Its
time-varying counterpart with nonconstant price
elastic-ity
e
D(t)
may also be used. Tenhunen [25] tested the
CED-function on Rautaruukki, a state-owned
Finnish steel manufacturing
"
rm, with yearly data
between 1990 and 1993. The estimated steel
quant-ities corresponded well with the realized
"
gures.
There are many possibilities to further re
"
ne and
develop our model. The riskiness of the business
environment can be recognized, e.g. by
Monte-Carlo simulations in the spirit of Kasanen et al. [13].
The impact of o
!
-balance sheet factors, such as
derivatives, is also relevant. Bessler and Booth [7]
have developed a bank model including derivative
securities. Another direction would be to concentrate
on techno-economic
"
rm planning, i.e. on
simulta-neous modelling of strategic decisions of the
"
rm and
calibration of its technical processes. Finally, our
"
rm-model could be extended to (multinational)
concerns, an important and worthwhile exercise.
Our results show that simple rounding of the
continuous solutions does not guarantee an optimal
mixed-integer solution in strategic
"
rm planning.
The problem with rounding errors in the cash
posi-tion deserves further attenposi-tion in future research.
Acknowledgements
Financial support from the Academy of Finland
is gratefully acknowledged.
References
[1] K. SoKderlund, R. OGstermark, A multiperiod"rm model for strategic decision support, Working paper, Asbo Akademi University, 1995.
[2] E. Kasanen, R. OGstermark, The managerial viewpoint in interactive programming with multiple objectives, Kyber-netes 16 (1987) 235}240.
[3] G.G. Booth, P.E. Kovesos, A programming model for bank hedging decisions, Journal of Financial Research 9 (1986) 271}279.
[4] H. Meyer zu Zelhausen, Commercial bank balance sheet optimization. A decision model approach, Journal of Banking and Finance 10 (1986) 119}142.
[5] A. Korhonen, A dynamic bank portfolio planning model with multiple scenarios, multiple goals and changing pri-orities, European Journal of Operational Research 30 (1987) 13}23.
[6] G.G. Booth, W. Bessler, W.G. Foote, Managing interest rate risk in banking institutions, European Journal of Operational Research 41 (1989) 302}313.
[7] W. Bessler, G.G. Booth, An interest rate risk management model for commercial banks, European Journal of Opera-tional Research 74 (1994) 243}256.
[8] D.W. Reid, G.L. Bradford, A"rm farm model of machin-ery investment decisions, American Journal of Agricul-tural Economics 69 (1987) 66}87.
[9] T. Westerlund, F. Pettersson, An extended cutting plane method for solving convex MINLP problems, Computers and Chemical Engineering 19 (Suppl.) (1995) S131}136. [10] T. Westerlund, H. Skrifvars, I. Harjunkoski, An extended
cutting plane method for a class of non-convex MINLP problems, Computers and Chemical Engineering 22 (1998) 357}365.
[11] R.A. Brealey, S.C. Myers, Principles of Corporate Finance, 4th Edition, McGraw-Hill, New York, 1991.
[12] R. OGstermark, Solving a linear multiperiod portfolio prob-lem by interior point methodology, Computer Science in Economics and Management 5 (1991) 283}302. [13] E. Kasanen, M. Zeleny, R. OGstermark, Gestalt system
of holistic graphics: New management support view of MCDM, in: A.G. Lockett, G. Islei (Eds.), Improving Deci-sion Making in Organizations, Lecture Notes in Econ-omics and Mathematical Systems, Springer, Berlin, 1989. [14] P. Asquith, D.W. Mullins, Equity issues and o!ering
dilu-tion, Journal of Financial Economics 15 (1986) 61}90. [15] P. Healey, K. Palepu, Earnings information conveyed by
dividend initiations and omissions, Journal of Financial Economics 21 (1988) 149}175.
[16] F. Modigliani, M.H. Miller, The cost of capital,
corpora-tion "nance and the theory of investment, American
Economic Review 48 (1958) 261}297.
[17] F. Modigliani, M.H. Miller, Corporate income taxes and cost of capital: A correction, American Economic Review 53 (1963) 433}443.
[18] J.B. Warner, Bankruptcy costs: Some evidence, Journal of Finance 32 (1977) 337}348.
[19] M.J. Gordon, Towards a theory of"nancial distress, Jour-nal of Finance 26 (1971) 337}348.
[20] M. Jensen, W. Meckling, Theory of the"rm: Managerial behaviour, agency costs, and ownership structure, Journal of Financial Economics 3 (1976) 305}360.
[21] J.B. Warner, Bankruptcy, absolute priority, and the pric-ing of risky debt claims, Journal of Financial Economics 4 (1977) 239}276.
[22] E. Altman, A further empirical investigation of the bank-ruptcy cost question, Journal of Finance 39 (1984) 1067}1089. [23] S.C. Myers, The capital structure puzzle, Journal of
Finance 39 (1984) 575}592.
[24] A. Kjellman, S. HanseHn, Determinants of capital structure: Theory vs practice, The Scandinavian Journal of Manage-ment 2 (1993) 91}102.
[25] L. Tenhunen, Supply functions and pricing behaviour in the theory of the"rm, University of Tampere. School of Busi-ness Administration, 1994, series A2. ISBN 951-44-3599-0.