CAPÍTULO III: MARCO METODOLÓGICO
3.5 RESULTADOS
3.5.1 Tabulación de encuestas
Chapter 2 showed several OR/MS approahes for supporting strategy proesses. How-
ever, thisdotoralthesisexplores theuseofSD modellingto supporttheinternationalisa-
tion proess throughstrategy rehearsal. Although there areother approahes to rehearse
strategy,suhasagent-basedmodels,protimpatofmarketstrategy,andbalanedsore-
ard(Dysonetal.,2007),IdeidedtouseSDmodellingbeausethisapproahhelpspeople
understand how strategies will perform over time, what might be wrong, and what kind
of intervention an be done (Kun and Moreroft, 2009). Also SD an helpompanies to
develop orretive ations from real-world feedbak with virtual performane to improve
strategi foresight (Kun andMoreroft, 2007a).
In traditional SD, the sequene of modelling ativities isas follows: theoneptualisa-
tion of the problemati situation; the development of model equations; the development
of a simulation to understand dynami behaviour; the evaluation of alternative poliies;
learning from the model and hoie of appropriate poliy; and nally, implementation
(Forrester, 1994). However, when thereis no onsensusamong deision-makersabout the
systemunderlying deisionsand thehypotheses(theory)generated forexplaininghowthe
system is not reahable beause deision-makers have dierent ideas (mental models) of
theproblem(reality)(KunandMoreroft,2009). Rouwette(2011)arguesthatfailitated
modellingisusefulinsupportingtheresolutionofstrategiissueswhenitisexpetedtoim-
prove ommuniation between deision makers, foster onsensus and reate ommitment.
Otherwise, models developed and interpreted by outside experts rarely hange the way
managers think about strategi issues (Senge and Sterman, 1992). Hene, the failitated
modelling proess an improve CEOs' abilities to view newsituations systematially and
dynamially(Moreroft,2007). ThisviewofSDmodellingonsidersthatCEOsthemselves
beomethe modellersof theirstrategy proess. Although thereisahighriskof modelling
a mental model that is aeted by mispereption of feedbaks (Sterman, 1989a), this is
thereal driver thatleads toimplement thestrategy.
Deision-makers argue that information about their rm's performane feeds bak to
modifyor reinforetheinitial deisions forimplementingastrategy (Kunand Moreroft,
2010). This feedbak proesses and its irular ausality has been a entral point in the
study of the dynamis of systems insoial sienes (Rahmandad et al.,2009). In system
dynamismodelling, aausalrelationshipisdenotedusinganarrowonnetinga
X
fatorthatis ausallyinuening
Y
(e.g.,X→Y
). Thisrelationship an be eitherpositive(+
)ornegative(
−
). Whentherelationshipispositive(/negative)means thataninreaseinX
leadsto aninrease(/derease) ofthelevelof
Y
. Whenthereareseveralsigniant ausalfators
X1, X2, . . . , Xn
inueningY
,thisrelationship is formallydenoted:However, when
X
andY
are aeted mutually (X
⇄
Y
), these relationships reate afeedbak loop. Feedbak loopsare dened asinterations among omponents ofa system
linkingations,events,orpieesofinformation (Sterman,1989a). Eventhoughomplexity
isarelativeterm,most omplexbehaviourusuallydependsonthenumberandthenature
of interations among the omponents of a system instead of on the omplexity of the
omponents themselves (Gharajedaghi, 2006). All dynamis arise from the interation
of two types of feedbak loops, positive and negative. Positive loops are self-reinforing
thereby generating exponential growth or deline. For example, if a rm inreases its
marketing ativities to gain market share in a loal market, its ustomers may respond
by buying more produts on sale. On the other hand, negative loops are self-orreting
therebyounteratinghange. Followingthelastexample,thehigherprodutdemandan
aet the ustomer servie quality when market dynamis are rapid relative to apaity
adjustment. In fat,foreasting errorsmayleadto a lakof apaity thatnallyan lead
to aderease inthenumberofustomers(See Figure3.2 ).
Figure 3.2: Example ofreinforing and balaningfeedbak loops
Within the SD eld, the onept of feedbak information plays a entral role in ex-
plainingthelinkbetween systemstrutureandbehaviour(Sterman,2000). Thefeedbaks
enhane the understanding of how individuals or groups make one or more deisions in
eah time period, based on the information available (Gary etal., 2008). However, dur-
ing the deision-making proess, managers might be aeted by insuient, unlear, or
oniting information about the results obtained from their deisions (Kun and More-
roft, 2007a). Aording to Moreroft (2007:5), `people's abilityto managetheir omplex
world an be improved by visualizing and simulating it'. In this sense, stok-and-ow
diagrams help to build mental models and represent thebasi knowledge strutures that
managersuseto make strategideisions (Kun andMoreroft, 2009). Inthisresearh,it
has onsidered a stok-and-ow diagram a set of strategi resoures onneted by arrows
denotingtheausalinuenesamongthevariablesextratedfromtheCEO'smentalmodel
of the analysed proess. Strategi resoures an be onsidering as asset stoks (Dierikx
andCool, 1989). Thesestoksaregraphiallyrepresentedasaretanglethataumulates
theirinowsandoutows. Aninowisdrawnasavalveonneted withanarrowthatbe-
ginsfromasoure andpoint atthe stok. Outowsarerepresentedwiththesame symbol
butthe owisoriginated inthestokandendsupinasink(soureandsinkaredrawnas
a loud)(Kun and Moreroft,2007a).
equations are employed to simulate the logial onsequenes, whih help rms to under-
stand the dynamis underlying the system. In this sense, Equation 3.11 formalises a
strategiresoure
i
(stok)(Rt
)astheinitialvalueoftheresoure(Ri(0)
)plustheintegral of investment inthis resoureovertime (ri(t)
):Ri(t) =Ri(0) +
Z
t
0
ri(t)dt
(3.11)Theurrent rateofaumulation
ri(t)
ofresourei
atthetimet
isafuntionoftheur- rentlevelofallexistingresouresaeting it(R1(t), R2(t), . . . , Rn(t)
,inludingexogenous fatorsdenotedgenerially asE(t)
.ri(t) =
f(R1(t), R2(t), . . . , Rn(t), E(t))
(3.12)Asset stokaumulations lie at the ore of all SD representations, whetheror not the
modellers oftheanalysishoosetomake itexpliit(Warren, 2004). Warren(2004)argues
that Stoks provide the onlymehanism by whih information andmaterial an be passed
forward through time. Unfortunately, the ognitive ability of managers visualising the
outomeof assetstokandlosedloopsystemsislimited, espeiallywhen aertaination
hasavaryingeetonanoutomedependingonitsintensitylevel(Sterman,1989b). These
non-lineardynamis arenot anunusualphenomenon. Indeed,manymanagerial situations
are aeted by onditions that hange depending on unertain fators. For example, in
Figure3.3therateof ustomerlossan impattheompanydierently depending onthe
levelof rise orfall oftheprodut demand andthe servie apaity.
As people's pereptions about a partiular system dier beause of the way in whih
a onsensusis reahed among individuals. Forrester (1994) postulates thatthelevel-rate-
feedbak struture of SD is indeed the fundamental struture of the real soial/physial
systems. However, withinorganisations, this ideaannot bevalidated againstthepart of
thereal world being modelled beausestakeholders areonstantly hangingthestruture
of the system. Regarding this view, models only show the pereption of the real world
at a speipoint of time. Thisaets managers' modesof theorising their ownstrategi
deisions (Colbert, 2004).
In this researh, simulation of SD models are based on the CEOs' pereption of their
own organisation's set of resoures. Hene, the simulations allowed the analysis of the
robustness of a CEO's assumptions about the pereived resoure system. This is not the
traditional SD approah of simulation. Nonetheless, onethere is agreement with CEOs
insupporting oneviewofstrategy,SD modellinganhelptoexplorethesolutionspaeby
rehearsing thepreferredstrategy through simulation. Inthis ase, SD models areusedas
transitional objets to failitatedialogue andtheexplorationoffutureinternationalisation
strategies. Then `simulations an help managers disover hidden pitfalls in strategy by
allowing them to rehearse resoure building(a task whih isdynamially omplex due to
interdependenies, time delays andnon-linearities' (Kunand Moreroft, 2009:198).