THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
EVALUATION OF THE BENEFITS FROM USING SYSTEM DYNAMICS
ON PERSONAL COMPUTERS FOR CORPORATE FINANCIAL MANAGEMENT
Prof. Albert HERREMANS Prof. Philippe WILMES
University of Louvain University of Louvain
Belgium Belgium
Info Center Manager President
IBM Internat. Education Center National Investment Corporation
La Hulpe - Belgium Brussels - Belgium
ABSTRACT.
DYNANO has been used for many years on mainframe computers. Like many other
applications programs and computer languages, it has been made available on
Personal Computers. Will that make System Dynamics easily available to many
potential users who never tried to use it on mainframes? Does one need to be
an Information System specialist to use DYNAMO on a PC?
These are the questions to which this paper intents to give an answer to.
Subsequently, we used DYNAMO on a PC for a very simple financial application,
and compared it superficially with the (already) traditional approach of
spreadsheets.
751
752
THE 1986 INTERNATIONAL: CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
INTRODUCTION.
Having used different versions of DYNAMO on an.IBM $/370 several years ago and
having had to stop any Systems Dynamics activity for a while, we thought it
would be interesting to verify if the recent technological evolution (hardware
and software) would allow new users to easily access the DYNAMO language.
In particular, we were interested by the use of DYNANO on PCs, and by its com-
parison with other PG tools, like the spreadsheet; we therefore took a very
simple example of financial management, not usually handled using System Dy
namics.
We shall first of all evaluate the DYNAMO language when used on a PC; and then
apply it to a simple financial model, and compare it to a well known
spreadsheet .
1. THE PROFESSIONAL DYNAMO (PD) FOR PC.
Having contacted Pugh-Associates Inc, in the USA, in October 1985, we decided
to order a copy of the Professional Dynamo ($500), the Professional Dynamo Plus
($2000) being too expensive for the limited experiment we wanted to perform;
the description of the additional functions of the PD+ versus those of the PD
did not give us the impression to justify the price differential.
The PD version 1.3 arrived from the USA in February 1986. Here are our obser-
vations; which we subdivided into comments on the package, comments on the
easiness of installation and use, and comments on the accompanying documenta-
tion.
THE PD VERSION 1.3 PACKAGE.
The package that came was made of:
2 diskettes: one contains the main program , the other one the supplemental
programs; the minimum PC configuration is 256K of RAM memory; the math
co-processor is not necessary, but used when it exists
a manual, that contains the User's Manual and the Reference Manual; this
manual is entitled "draft version 1.3", showing that it is not the defini-
tive manual
the Reference card is said to accompany, but does not!!
EASINESS OF INSTALLATION AND USE:
installing the PD either on diskettes or on a hard disk is rather simple;
the User's Manual explains how to do it; the user has the right to make 2
copies of the original diskettes; the procedure used combines programs from
both diskettes into a working diskette, and decreases a counter for every
installation, so that the user could not make more than 2.
Particularly interesting, and not existing (yet) for many PC SW nowadays,
is the possibility to "uninstall" PD from either a diskette or a hard disk:
when something goes wrong during the PD installation, or when the user
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
whishes to use a hard disk instead of a diskette (or vice versa) and has
already installed the program twice, he/she may "uninstall" it, which in-
creases the counter by one unit again!
Our only remark about the installation is that users generally place the
software diskette in disk drive A: and the data diskettes in drive B:
meanwhile the PD procedure suggests that the PD working diskette should be
placed in drive B: and the data diskette in drive A:
the command menu (and subcommands menus) works pretty much like the LOTUS
one(s): move the cursor on the desired command and press the Enter key, or
just press the first letter of the desired command name; PD is clever enough
to anticipate what the next command will be, and automatically moves the
cursor to it: after editing a model, it is usual to compile it, then simu-
late and look at the results!
five commands are offered on the PD main menu:
* EDIT, used to enter or modify a model, or a file with the desired output
specifications (tables and graphs)
* COMPILE, used to compile the program and check syntax and missing
specifications; if the compilation may not be achieved correctly, the
model is automatically edited for modifications
+ SIMULATE, used to run the model once or several times, a subcommand
giving the user the opportunity to change parameters for more runs
+ REPORT (sometimes referred to as VIEW in the litterature), shows on the
screen or prints.on the Graphics printer the desired output
* DOCUMENT appears on the menu, but is not available in version 1.3!!
* HELP
A tutorial is contained in the User's Manual, which is very good at helping
the beginner to get trough the different commands and PD phases.
the utilization of the PD program itself proved to be very easy, at least
for somebody already used to work with a PC; but the menu approach would
not create any problem to a PC beginner, in our opinion.
the SIMULATE command was a bit confusing, because it does not simulate im-
mediately: it offers the possibility to change some parameters, even when
coming directly from COMPILE
REPORT produces tables and graphs which are not different from the ones
known for many years now on the mainframe versions of DYNAMO; PD does not
take advantage of the PC possibilities to offer a clearer output, which is
disappointing. The PD+ version (4 times more expensive) is said to offer a
better output, but we did not get a chance to test it
PD files: PD creates different types of files at every step of its use:
* EDIT is used to create or update *.DYN files (the models) and/or *.DRS
files (display specifications: print and plot requirements)
* COMPILE reads *.DYN files and creates 3 types of files:
and *.INS which will be used by SIMULATE
SMT, *.DAT
at SIMULATE uses the files created by COMPILE, and produces *.RSL files
containing the results of the simulation; it also creates *.SIT files
when using the Preserve/Resume possibility (to start a new simulation
from where the previous one stopped)
* REPORT uses the *.DRS files (display specifications) and applies their
. content to the *.RSL files to produce tables and plots
* DOCUMENT is not available in this version of PD
753
754
8.
‘THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
only models (#.DYN) and,display specifications (*.DRS) files may be edited,
which prevents the user from developping his/her own pieces of programming
(e.g. plotter output, better table presentation, ..).
THE DOCUMENTATION.
The manual(s) is really a draft; figures are not clean, and the PD Reference
Manual part is far from being complete: only chapters 3,/4, 6, 7 and 8 exist;
the last chapters' pages are mixed up!! Not easy to find your way. Error mes-
sages are not explained yet in this version.
THE PD V 1.4 PACKAGE.
In April 1986, we received the PD version 1.4: 2 new diskettes, and a new (draft)
of the User's Manual and Reference Manual. At the same time, Pugh-Associates
Inc was asking to send back the original diskettes of version 1.3
From the accompanying etter, we understood that version 1.4 includes new fea-
tures, and that the documentation is now 90% complete.
At first, we were unable to install version 1.4 following the instructions of
the Manual; having contacted Pugh-Associates Inc, it seemed that this was hap-
pening with some PCs and not with others! Thé problem is in relation with the
«protection scheme of the package: the SIMULATE module is hidden from the DOS,
and sometimes the commands issued from a Menu do not find it! To make it more
positive, we experimented here how interesting the "uninstall" procedure is!
While waiting for a solution (or just an answer about what to do in order to
use the PD V 1.4), we had a quick look at the new (draft) Manual: it shows indeed
a
real progress in the amount of available documentation, but the last chapters
(altough different from the 1.3 version) are consistently mixed up!! Page 10-4
is followed by the end of Chapter 15, of which the first page is followed by
the last one of Chapter 14, etc... until page 10-6 and 10-5 ends the Manual.
Beginning of June 1986, another call to the USA allowed us to understand that
the problem we had would be solved if we were using DOS 3.0 or above! And it
does, indeed! But how many weeks lost because of this!
EASINESS OF INSTALLATION AND USE:
The PD V 1.4 comes like V 1.3, on 2 diskettes; its installation does not
offer any difficulty; the user has the right to only one copy this time;
although the messages on the screen during installation time say one may
install 2 copies, the counter goes down to 0 after the very first instal-
lation (2-1=0!). In conformity with the manual...
the modules are about the same as in V 1.3, with the following differences:
* the *.DYN file does not have any more to specify "print" and "plot"
requirements; all variables listed in the SAVE statement(s) are auto-
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986. 755
matically candidates for prints and plots; this means that *.DRS are
not necessary any more
© REPORT is replaced by VIEW, already mentioned in the V1.3 Manual; VIEW
offers much more flexibility than REPORT was:
- PLOT selected variables: the user chooses among the complete list
of saved variables; but there is still no plotter output! Graphics
printers will only produce a hardcopy if the GRAPHICS command was
called from DOS; one can plot several or all variables using the
same scale
— TABULATE: either all variables, or a selection; in this case only,
the table is produced in the order of the selected variables
* DOCUMENT exists this time! but on a second diskette (created during
installation); but when the command is invoked, PD says it does not find
the module, and does not prompt the user to place the appropriate
diskette in drive B: !! But the module is interesting: it allows the
user to re-order his/her model in a.much clearer and readable way; it
helps the user convert models written for previous DYNAMO versions; it
creates *.DOC files, and sometimes *.DEF files (for definition files)
* HELP: is context-sensitive; the COMPILE module has no Help; the SIMULATE
module has one, but when invoked, the message "RUN.HLP missing" ap-
pears..... In fact, it is not copied on the working diskette during
installation, because of insufficient space on the diskette...
3. when in EDIT, PD sends the user back to DOS when there are errors in the
model (eg. "failure to recognize end of line"), without explanation.
4. no COMPILE error messages in the V 1.4 manual; one can read the errors at
the bottom of the screen, using Alt-N to look at the next error message,
but how interpret it without the appropriate Manual pages? In particular,
having used EXP (for Expenses) as a variable name, no clear message said
this was a protected name (for Exponential); the error messages were dif-
ficult to interpret.
5. no way to plot other plots than line charts (eg, bar or pie charts)
6. no way to write kind of an EXEC to reproduce a series of commands in a
certain order.
7. in PD V 1.4, the print a PLOTfunction does not work: one has to use the
Prt-Ser.
8. in SIMULATE, one can save different runs under different names; one can
re-start from the end of a certain simulation, and continue it further using
the Preserve/Resume function, which uses *.SIT files.
9. in VIEW, views are numbered and may be called back just by specifying their
number; but only as long as the VIEW is active; no possibility to save the
views under a name ‘for later usage
10. error messages in the 1.4 manual: only a few for EDIT; none for the other
modules.
GENERAL COMMENTS AND CONCLUSIONS ABOUT PD.
Our limited experience with Professional Dynamo, V 1.3 as well as 1.4, does not
allow us to formulate a definitive opinion, nor give a valuable advice on when
and why use it. Nevertheless, we believe this PC software really opens the way
to Systems Dynamics for a much larger number of potential users, even if they
are not using the PC yet for other applications.
756
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
We are waiting for Version 1.5 (or 2.07), whatever the name and number; it
probably will be the definitive one, with all functions and the complete doc-
umentation. Its use will avoid some of the little problems we almost pioneered
with the existing versions.
No doubt about the easiness to install and learn how to use PD versus the in-
stallation and learning process in order to use DYNAMO on a mainframe computer.
Looking at performances, one can already learn and develop medium size models
on a usual PC, with 256K of memory and 2 floppy diskette drives! Of course,
the use of more powerful PC models, in terms of memory size and instructions
per second, allows the user to build much bigger modeis and process them with
a still acceptable response time.
Our only regret is that PD does make full use of the PC possibilities which are
usually used in all PC software, mainly in graphics (access.to more than just
line charts) and output (plotter output, and better control of all outputs:
tables and graphs).
But this is said to exist in the PD+ package, which is unfortunately 4 times
more expensive!
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986. 757
2. DYNAMO OR FINANCIAL PLANNING TOOLS FOR FINANCIAL MANAGERS?
A base model (Figure 1 on page 8) was written with PD, based on the following
assumptions:
* a company has a revenue of 50 (whatever the monetary unit) and expects an
annual increase of 25% over the 5 years of its strategic planning period
* its manufacturing cost is 19, and is likely to growth at 25% a year
* general expenses are 17, and would growth at 33% a year
* R&D expenses represent 4% of the revenue
* tax rate is 50% of EBIT (earnings before income taxes)
758
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
—
* revenue
*
rev. j+dt*revchg. jk
n rev=irev
cirev=50
x revchg.kl=rev.k/4
cost
19
costchg.kl=cost .k/4
expenses
exp. jdt gexpchg. jk
gexpchg. kl=gexp.k/3
research & development
1
a
¢
r
*
*
1
n
c
x
*
*
a rdexp.k=rev.k*.04
*
* earnings before tax
#
a ebit.k=rev.k-cost .k-gexp.k-rdexp.k
=
* provision for taxes
*
a tax.k=ebit.k*.50
*
* net profit
*
a
x
profit .k-ebit.k-tax.k
save rev,cost ,gexp,rdexp,ebit,tax,profit
save revchg,costchg, gexpchg
spec dt=1/length=5/savper=1
Figure 1. PD base model
‘The same model was also developped in LOTUS123.
There is no difficulty to use the PD for discrete financial planning as with
existing and widely used tools such as LOTUS, etc... The equations have very
simple forms (levels and rates).
The only problem is to put the value of the computation interval DT, which is
specified in the SPEC equation, at the value of the period under consideration.
Moving from one unit time interval (such as year) to another time interval (such
as months or quarters) must be done taking into consideration the discounted
effect of another time unit. Example: 12% per year ; monthly rate is NOT 1%..
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
A first simulation of MOD1 shows an increasing difference in the results between
DYNAMO (Figure 2 on page 9)'and LOTUS ( Figure 3 on page 9)!!
A special attention must be put on the value of the relative error parameter
(REL-ERR), which is set by default to 10E-3; this REL-ERR value enables the
computer to execute the integration process with a variable step size under the
Runge-Gutta method. Variable’ step integration generates an unfortunate dis-
crepancy between the discrete financial planners results and the DYNAMO ones.
One has to give REL-ERR the value 0 to integrate with the Euler method, and so
obtain results which are comparable with the discrete models. There is no other
way of saving the REL-ERR value of 0 than entering its value in the SPEC
statement of the model; otherwise, the default value is used everytime the
SIMULATE function is talled!!
modl.rsl1 - Page 1 MOD1 Runge-Kutta
TIME 0. my 2. 3. 4, 5.
REV 50. 64.2 82.423 105.82 135.86 = 174.42
cosT 19. 24,396 = 31.321 40.211 51.626 66.28
GEXP 17, 23.724 33.096 = 46.172 64.412 89.859
RDEXP 2. 2.568 3.2969 4.2328 «= 5.4343 6.9768
EBIT 12, 13.512 14.709 15.204 14.385. 11.305
TAX 6: 6.756 7.3546 7.6019 7.1925 5.6525
PROFIT 6. 6.756 7.3546 7.6019 7.1925 5.6525
REVCHG 12.5 16.05 20.606 =. 26.455 33.964 43.605
COSTCHG 4.75 6.099 7.8302 10.053 12.906 16.57
GEXPCHG 5.6667 7,908 11.032 15.391 21.471 29.953
Figure 2. Model 1 - PD - Runge-Kutta
time ° 1 2 3 4 5
rev 50.00 62.50 78.13 122.07 152.59
cost 19.00 23.75 29.69 46.39 57.98
gexP 17.00 22.67 30.22 53.73 71.64
rdexp 2.00 2.50 5P43 4.88 6.10
EBIT 12.00 13.58 15.09 17.07 16.86
tax 6.00 6.79 7.55 8.54 8.43
profit 6.00 6.79 255 8.54 8.43
revchg 12.50 15.63 19.53 30.52 38.15
costchg 4.75 5.94 7.42 11.60 14.50
gexpchg 5.67 7.56 10.07 17.91 23.88
Figure 3. Model 1 - LOTUS
A second simulation was made, using the Euler method; the results are the same
as with LOTUS (Figure 4 on page 10).
759
760 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
modi.rsl - Page 1 MOD1 Euler
TIME o. 1. 2. 2 4. 5.
REV 50. 62.5 78.125 97.656 122.07 152.59
cost 19. 23.75 29.688 += 37.109 46.387 57,983
GEXP 17. 22.667 © 30.222 40.296 = 53.728 = 71.638
RDEXP 2, 2.5 3.125 3.9063 4.8828 = 6.1035
EBIT 12. 13.583 15.09 16.344 17.072 16.863,
6. 6.7917 7.5451 8.1722, 8.5362 88.4316
PROFIT 6. 6.7917 7.5451 8.1722 8.5362 8B. 4316
REVCHG 12.5 15.625 19.531 24.414 30.518 = 38.147
COSTCHG 4.75 §.9375 724219 9.2773, 11.597 14.496
GEXPCHG 5.6667 7.5556 10.074 13.432 17.909 23.879
Figure 4. MOD 1 - PD - Euler
The next question is: what if the annual growth rates of revenue, costs and
expenses are not constant? This is easy to tackle in a LOTUS model (Figure 5);
PD may address the same problem using tables (Figure 6 on page 11 shows the
model and Figure 7 on,page 12 its results). The results are the same, if one
uses the Euler integration method, of course...
time ° 1 2 3 4 5
rev 50.00 60.00 72.00 © 82.80 95.22 114.26
cost 19.00 23.75 28.50 34.20 39.33 45.23
gexP 17.00 22.10 27.63 34.53 41.44 49.73
xdexp 2.00 2.40 2.88 3.31 3.81 4.57
EBIT 12.00 11.75 12.99 10.76 10.64 14.74
tax 6.00 5.88 6.50 5.38 532, 7.37
profit 6.00 5.88 6.50 5.38 5.32 137
revchg 12.50 15.00 18.00 20.70 23.81 28.57
costchg 4.75 5.94 7.13 8.55 9.83 11.31
gexpchg, 5.67 wae 9.21 11.51 13.81 16.57
trev 0.20 0.20 0.15 0.15 0.20 0.20
tcost 0.25 0.20 0.20 0.15 0.15 0.25
tgexp 0.30 0.25 0.25 0.20 0.20 0.20
Figure 5. LOTUS model with variable growth rates
‘THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986. 761
revenue
rev.k=rev. j+dt*revpet. jk
rev=50
revpct .k1=
revehg.k=table(trev,time.k,0,5,1)
trev=.20/.20/.15/.15/.20/.20
cost
ost. jtdt*costpet. jk
costpct .kI=cost .k*costchg.k
costchg.k=table(tcost, time.k,0,5,1)
teost=.25/.20/.20/.15/.15/.25
expenses
gexp. k=gexp. jtdtgexppet. jk
gexp=17
gexppet .kl=gexp.k*gexpchg.k
gexpchg.k=table(tgexp,time.k,0,5,1)
tgexp=.30/.25/.25/.20/.20/.20
research & development
rdexp.k=rev.k#.04
earnings before tax
ebit.
=rev.k-cost -k-gexp.k-rdexp.k
provision for taxes
tax. k=ebit.k*.50
actual net profit
et ee ee ee eee eee ee ee ee
profit .k=ebit.k-tax.k
save rev,cost,gexp,rdexp,ebit ,tax,profit
save revpct,revchg,costpct ,costchg, gexppct , gexpchg
spec dt=1/length=5/savper=1
Figure 6. PD model with tables for variable growth rates
762
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
modit.rsl - Page 1 MODIT Euler
‘TIME 0. 2. 3. 4. 5.
REV 50. 72. 82.8 95.22 114.26
cosT* 19. 28.5 34.2 39.33 45.23
GEXP 17. 27.625 34.531 41.438 49.725
RDEXP Be 2.88 3.312 3.8088 = 4.5706
EBIT 12, 12.995 10.757 10.644 14.739
TAX 6. 6.4975 5.3784 = 55.3218 = 7.3695
PROFIT 6. 6.4975 5.3784 5.3218 = 7.3695
REVPCT 10. 10.8 12.42 19.044 22.853
COSTPCT 4.75 5.7 5.13 5.8995 11.307
GEXPPCT 5.1 6.9063 6.9063 8. 2875 9.945,
Figure 7. PD model with tables for variable growth rates
Although the PD has the capability to simulate all functions of the traditional
discrete financial planners, its use is sometimes more tedious and cumbersome.
Its main advantage therefore must be viewed in the enhanced possibility to
connect (simulated) discrete financial planning models with general continuous
corporate models such as strategic development ones.
1
In the next instance, we will add a few equations which will connect the tra-
ditional financial planner with a control equation relating the corporate fi-
nancial results (profit after tax) with the R&D policy. The R&D expenses are
viewed as creating after a given time lag new products and services which gen-
erate additional sales. The control policy of increasing the R&D effort takes
into consideration the discrepancy between a given desired profit and the actual
results.
In the design of the different continuous equations, one must remember ‘that the
value of the computing interval DT which has been set in the financial model
to the value of the period under review, must as a rule of thumb not be smaller
than the value of 1/2 the smallest delay.
So, let us now compare the 2 modelling systems if a feed-back is introduced in
the model. Let us say that the company adopts a R&D policy that says that if
the net profit after tax does not represent 15% of the revenue, the company has
to invest more in R&D expenses: 80 cents for every § that is missing in the net
profit to reach the target! This would have an impact on the revenue, that would
growth faster than planned, but with a delay of 3 years, because of the time
Jag between deciding to invest more in R&D and the possible effect of this de-
cision on the revenue. Costs are also growing when producing more to sell more!
General expenses are supposed to grow at their normal rate, without being im-
pacted by this additional business.
This is done in MOD2 (Figure 8 on page 13 shows the model, and Figure 9 on page
14 its results); the equivalent LOTUS model is shown in Figure 10 on page 14.
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986. 763
revenue
rev.k=rev, j+dt*(revehg. jkteard. jk/10)
rev=50
revchg -kl=rev.k*revpct .k
revpct .k=tabhl(trev, time.k,0,5,1)
trev=.20/.20/.15/.15/.20/.20
cost
cost .k=cost. j+dt* (costchg. jteard. jk/40)
cost=19
costchg.kl=cost .k*costpet -k
=tabhl (tcost, time.k,0,5,1)
5/.20/.20/.15/.15/.25
expenses
gexp.k=gexp. jtdt*gexpchg. jk
7
gexpchg.kl=gexp.k*gexppct .k
gexppct .k=tabhl(tgexp,time.k,0,5,1)
=.30/.25/.25/,20/.20/.20
research & development
rdexp.k=(rev.k*.04)+laard.k
earnings before tax
ebit .k=rev.k-cost .k-gexp.k-rdexp.k
provision for taxes
tax.k=ebit.k*.50
actual net profit
profit .k=ebit.k-tax.k
desired net profit
dprofit .k=rev.k*.15
expected additional R&D expenses rate
eard.kl=z*(dprofit.k-profit.k)
22.8
actual additional R&D expenses rate
-kelaard. j+dt*aard. jk
laard=0
BEN RESON FEED FEED FRED FRED RP AD EPROM DEAS PONS HEATON BREE
save rev,cost,gexp,rdexp,ebit ,tax,profit ,dprofit,eard,aard,laard
save revpct,revchg,costpct ,costchg, gexppct , gexpchg
spec dt=1/length=5/savper=1/rel_err=0
Figure 8. Model 2 - PD - with feedback
764 THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
mod2.rsl - Page 1 model 2
TIME 0. 1 2 3. 4 5
REV 50. 60.12 72.44 83.739 96.992 117.27
cost 19 23.78 28.61 34.44 39.779 45.966
GEXP v7 22.1 27.625 34.531 41.438 49.725
RDEXP 3.6048 5.2976 6.9496 8.6797 12.451
EBIT 12 10.635 10.907 7.818 7.0955 9.1281
TAX 6 5.3176 5.4537 3.909 3.5478 4.5641
PROFIT 6 5.3176 . 5.4537 3.909 3.5478 = 4.5641
DPROFIT 7.5 9.018 10.866 12.561 14.549 17.591
EARD 1.2 2.9603 4.3298 © 6.9215. 8.8008 —:10.421
AARD 1.2 1.2 1.2 1.2 2.9603 4.3298
LAARD oO 4.2 2.4 3.6 4.8 7.7603
REVCHG 10 12.024 10.866 12.561 19.398 23.454
REVPCT 2 E +15 a6 2 2
COSTCHG 4.75 4.756 5.722 5.166 5.9669 11.492
COSTPCT 25 a2 eB: 2S: iS 25
GEXPCHG 5.4 5.525 6.9063 6.9063 8.2875 9.945
GEXPPCT oa +25 «25 ot 12 +2
Figure 9. Model 2 - PD - with feedback
—
time oO 1 2 3 4 5
rev 50.00 60.12 72.44 83.74 96.99 117.27
cost 19.00 23.78 = 28.61 34.44 © 39.78 45.97
gexp 17.00 22.10 27.63 34.53 41.44 49,73
rdexp 2.00 3.60 5.30 6.95 8.68 12.45
EBIT 12:00 10.64 10.91 7.82 7-10 9.13
tax 6:00 5.32 5.45 3191 3.55 4156
profit 6.00 5:32 5.45 3.91 3.55, 4.56
dprofit 7.50 9:02 10.87 12.56 14.55 17.59
eard 1.20 2.96 4.33 6.92 8.80 10.42
aard
eard 1.20 2.96 4.33 6.92 8.80 10.42
aard 1.20 1.20 1.20 1.20 2.96 4.33
laard 0.00 1.20 2.40 3.60 4.80 7.76
trev 0.20 0.20 0.15 0.15 0.20 0.20
tcost 0.25 0.20 0.20 0.15 0.15 0.25
tgexp 0.30 0.25 0.25 0.20 0.20 0.20
Figure 10. Model 2 - LOTUS - with feedback
THE 1986 INTERNATIONAL CONFERENCE OF THE SYSTEM DINAMICS SOCIETY. SEVILLA, OCTOBER, 1986.
CONCLUSIONS
The use of Professional Dynamo, although it is not available yet in its defin-
itive version, proved to be very easy and powerful; some minor improvements
could be suggested, but we intend to wait for the next version before making
any suggestion of any kind.
It certainly opens the use of Systems Dynamics to a lot of potential users, they
be already PC users or not.
For financial management, the use of LOTUS or Professional Dynamo give the same
results for the limited experiments we made with both systems. Financial man-
agers are certainly more used to think in terms of discrete models, but when
wishing to integrate their models into a larger one, or when wishing to expand
‘the use of their models to more complex relations, Systems Dynamics might be
mach better than discrete approaches.
As an example, our Model 2 could be improved by including a set of equations
that would suggest an optimum R&D policy, bringing progressively the actual
profit at the level of the desired profit; of course, this would be a function
of our simplification, i.e. that R&D is the only factor that allows to grow
faster than the normal trend, inducing more revenue and more profit!
765
CC BY-NC-SA 4.0