EVALUATING THE ROLE OF INDIVIDUAL PERCEPTION IN IT

OUTSOURCING DIFFUSION

An Agent based Model

Marco Remondino, Marco Pironti

e-business L@B, University of Turin, Cso Svizzera 185, Torino, Italy

Roberto Schiesari

Faculty of Economics, University of Turin, Cso Unione Sovietica, Torino, Itay

Keywords: Innovation diffusion, Simulation, Agents, Data mining, Management, Strategic planning, Perception.

Abstract: The decision to adopt innovations has been investigated using both international patterns and behavioural

theories. In this work, an agent-based (AB) model is created to study the spreading of innovation in

enterprises (namely, the adoption of Information Technology outsourcing). The paradigm of AB simulation

makes it possible to capture human factors, along with technical ones. This makes it possible to study the

influence of perception, and the resulting bias. This work is focused on small and medium enterprises

(SME), in which a restricted managing pool (sometimes just one person) decides whether to adopt a new

technology or not, and bases the decision mainly on perception.

1 INTRODUCTION

A technological development often does not seem to

be matched by a parallel movement towards

adopting and exploiting the opportunities it offers,

through economic organizations, especially for

SMEs. Therefore, an important goal is to outline the

methods through which firm decision makers choose

to start innovative investments and identify, at the

same time, the criteria with which this choice is

made. This was first discussed in the studies carried

on by Rogers (1983). The author outlined five

features of innovation (relative advantage,

compatibility, complexity, observability and

triability), which influence its adoption, and

consequently affect its circulation rate. Besides, the

spreading of IT innovations cannot be modelled

simply by taking into account pure technical issues;

sometimes the forecasts made with a statistical - or

process-based model, which considers just the

engineering or financial aspects (i.e.: production

process, activity decomposition, scale economies,

product life cycle and so on), prove to be inaccurate

or even wrong. This was the case with cellular

telephones, air-cooling systems, personal computers

and other fields in which some external influences

have led to results which are very different from the

initial pre-dictions obtained through prevision tools.

The purpose of this work is to examine the

circulation of an emerging technical innovation:

using IT services in Application Service Provider

(ASP) mode. The spreading of this technology

should not be affected by phenomena like fashion, as

with cellular phones, and its adoption is not

considered as a status symbol, although, there are

some other influences which can-not be taken into

account by a process-based or statistical model, i.e.:

the effects of individual perception about savings,

strategic impacts and so on. Therefore, a very

important bias seems to be individual perception,

and, in turn, the social links that exist in the network

connecting the enterprises, from which the

perception mainly originates and spreads.

2 PERCEPTION ISSUES

Proper perception of the benefits granted by the

adoption of an innovation is a crucial element in

155

Remondino M., Pironti M. and Schiesari R. (2009).

EVALUATING THE ROLE OF INDIVIDUAL PERCEPTION IN IT OUTSOURCING DIFFUSION - An Agent based Model.

In Proceedings of the 11th International Conference on Enterprise Information Systems - Software Agents and Internet Computing, pages 155-158

DOI: 10.5220/0001856701550158

 SciTePress

order to understand the circulation potential and the

growth rate for that innovation.

By following a strictly rational approach, it often

proves difficult to understand the reasons for

development difficulties and also for the delays in

the circulations of innovative goods and services.

It may therefore prove useful to analyse the

reasons that may distort perception of the advantages

resulting from the outsourcing of IT services in the

ASP mode by SMEs, which are on the other hand

the parties which are likely to benefit from the

greatest advantages provided by the broadening in

the offer of such services.

The reduction in initial investments, the easier

and less onerous opportunity to benefit from the

improvement of technology, as well as the greater

certainty about costs and their manifestation in time

should be strong motivations for the spreading of

such services among smaller enterprises, once the

infrastructure prerequisites related to the

performance and reliability of data transmission

networks and to the existence of an offer market

capable of creating competition have been

established.

However, we should take into account the fact

that Italian SMEs are late in IT-related expenses and

investments.

Indeed, a recent research published in Italy

(Confcommercio, 2007) stresses how 53% of IT-

related expenses and investments come from

enterprises with over 500 employees, or even 76%,

if we also take into account enterprises with over

100 employees.

IT-related expenses and investments are

therefore restricted to a limited number of larger

enterprises than the over 5 million Italian enterprises

with a lower number of employees which have

invested little or no money, whereas it is known that

the ICT-productivity growth ratio is extremely high.

This ratio is obvious, not only in industry but also in

trade and services. The European committee itself

has requested that, at a micro-enterprise level, the

element represented by shared organisations be

developed.

SMEs’ requirement to benefit from IT services

that are fit to face existing competition challenges is

therefore a priority for their very existence.

However, the growth rate in resorting to

outsourcing by SMEs has proved lower than

expectations up to very little time ago.

In order to try and understand the reasons that

might distort the perception of outsourcing

advantages, it may be helpful to mention the fear

experienced by smaller enterprises of ending up

depending from opportunistic behaviours that might

be adopted by the suppliers of such services, and

which smaller enterprises fear they may be more

vulnerable to, owing to their lower bargaining power

and more limited understanding of the alternatives

offered by the markets (information asymmetries).

Indeed, opportunistic behaviours represent one

of the reasons for market failure that cause

enterprises to internalize the activity, in order to gain

some kind of insurance. The existence of such

behaviours causes transaction costs to increase and,

in the event of a high level of asset specificity,

encourages enterprises to maintain such activities

within the company or to internalise them.

This issue is related to the problem of accurately

assessing switching costs to be borne in the event

that the outsourcing IT services supplier is changed,

which are even more difficult to evaluate by an

enterprise that lacks specific competence.

Another problem which significantly hinders

outsourcing is related to the confidentiality of

information and data contained in the enterprise

information system and hence depends on the

reliability of the supplier of the service involved in

safeguarding the data themselves, both during the

course of the cooperation agreement and in the event

of termination of services, also if this is related to

controversies that may have arisen between the

enterprise and the supplier.

A SME may perceive this risk both because it

lacks the resources and skills required to asses the

reliability of the systems and process implemented

by the privacy supplier and because of the belief,

especially strong in Italy, as to the non-existence of

a prompt and efficient legal protection in the event

that one’s privacy rights are infringed.

The bounded rationality characterising the above

mentioned processes is amplified by formalised, but

above all informal, relations among the parties

involved, within an enterprise, but also among

parties belonging to different organisations.

The “confidence” and “trust” problems

characterising many SMEs’ approach to IT

outsourcing represent a restraint to horizontal

communication between enterprises, which on the

other hand could represent a major innovation

propagation factor.

3 MODEL DESCRIPTION

A computational model is built, by using Java

coding. It consists of a user interface, in which the

initial simulation values are set. According to

ICEIS 2009 - International Conference on Enterprise Information Systems

156

defined rules, the model computes these data, in

order to give final aggregate results.

The main purpose is to analyze how individual

perception can bias real and objective evidences. A

number n of macro-agents is built, representing the

SMEs; each of those is composed of a random

number of agents (with a min/max range),

representing the managing pool.

The agents belonging to the same enterprise are

connected by a social network; the links feature a

random weight, with a modifiable threshold. The

network has the purpose of transmitting the

information about the technology. Some of the

agents can be connected to others, belonging to

different enterprises, thus spreading their opinion

over the network. The perception is derived from

real savings allowed by the new technology, but not

only that; there could be a bias (non-perfect

perception), representing other factors, like

trustfulness towards the company supplying the IT

outsourcing facilities, personal inclination towards

IT and so on. The agents spread their own

perception to the others connected to them and this

will be modified ac-cording to the weight of the

specific individual link. Only agents are directly

connected, not the enterprises; agents represent

persons, which can be connected some-how (e.g.:

friendship), while enterprises are considered

“connected” among them if and only if the people

belonging to them know each other. Since the focus

is how social relations and individual perception

influence the spreading of IT outsourcing in SMEs,

the model, through repeated execution, shows the

diffusion over time, when changing one parameter at

a time, while leaving the others unchanged (coeteris

paribus).

3.1 Parameters and Iterations

Here follows a list of the main parameters in the

model:

a. Total number of enter-prises

b. Average number of agents per enterprise

[managing pool]

c. Percentage of SMEs that already adopted

d. Costs for the innovation (una tantum and

variable cost)

e. Average saving allowed by the new technology

f. Perception BIAS [double, perfect, inverse,

random perception]

g. Satisfaction Threshold (ST) for adoption

[range: 0 to 1]

h. Requested majority to adopt the IT outsourcing

in the enterprise.

These parameters can be changed at the

beginning of each simulation, in order to create a

scenario. Here follows the sequence of steps

performed by the simulation:

1) The first step is the adoption of the new

technology by a number of enterprises

2) This has installation costs, and brings a

positive or negative saving

3) These will be perceived by the agents in the

enterprise, filtered by a personal BIAS

4) Each agent randomly spreads her own

perception, and the agents linked with her will form

their own opinion, according to the average of the

received opinions

5) According to the ST, the agents will

individually decide whether to adopt or not

6) Once all the agents in an enterprise form their

own opinion, according to the re-quested majority,

the SME will adopt the new technology or not. Loop

from 2).

There is not a metaphorical correspondence

between the steps and time units. It’s possible to

realistically think about each step as one month time.

4 RESULTS AND COMMENTS

Several experiments have been done on the model,

by changing the parameters in order to mimic

different scenarios. Two are the core parameters

studied in this work that affect the diffusion over

time: a different connection level and a different

perception level. In particular, since we compare the

results with empirical data coming from a localized

market (namely, Italian one), sooner or later there

will be a saturation level, that’s a threshold beyond

which it’s not possible to go (i.e.: a limited number

of agents can be reached by an innovation, since a

limited number of enter-prises exist in a market). So

we aspect an highly connected network to generally

increase the diffusion speed over time. We think to

horizontal connections (e.g.: clubs of managers). In

those situations, people spread their opinion among

others working in the same field or, at least, with the

same job and position, making their message more

effective. The saturation threshold is represented by

the parameter called “Total number of enterprises”

that multiplied per “Average number of agents per

enterprise” gives a good approximation of the total

number of agents in the simulation (not exact, due

statistical distributions and consequently to sampling

effects). On the graphs, the total adoptions versus

time step are shown. A total of 500.000 macro

agents (enterprises) are set in the model and an

EVALUATING THE ROLE OF INDIVIDUAL PERCEPTION IN IT OUTSOURCING DIFFUSION - An Agent based

Model

157

average of 5 is the managing pool hypnotized. In

order to simulate a realistic situation, 20% of

enterprises are considered as already reached by the

new technology at time zero. The costs for the

innovation and the average savings allowed by the

new technology are deduced by interviews made

with a sample of Italian SMEs. The requested

majority is set at 51%, since we imagine all the

members of the managing pool having the same

decisional power. With a low perception index and a

loosely connected network, after one simulated year,

from the initial 100.000 enterprises that already

adopted IT outsourcing, we have that 1.399 moved

to the new technology, that’s a tiny 1,4% increase.

Same poor figure for the following years: 1,1% for

the second, 1,2% for the third, 1,6% for the forth and

finally a slightly better 2,5% for the fifth. The slow

increase, even with the high savings allowed by

adopting the new technology, are to be explained

taking into account the low perception index. The

bias that the perception introduces in the model is

enough to make a very good deal to look like an

average or poor one for many agents in the model.

Besides, the loosely connected network prevents the

agents to spread the good news about their savings

to many others and so, after 5 simulated years, less

than 10% new enterprises stepped to the new

technology, from the original 100.000.

The same values were for the 2nd experiment,

but for the level of connection for the network, now

an highly connected one. This is an unrealistic

scenario, since it would mean that every manager

knows almost all the other ones. The perception in

this experiment is still very low, meaning that even a

big advantage is not perceived as such.

The results are now quite different. The first year

brings an increase of 21%, the following simulated

years also have a very good trend: in fact, the second

year brings an increase of 15%, the third of 12%, the

forth of 8% and the last one of 10%.

The 3rd experiment is about a loosely connected

network, as that in the 1st experiment, but with

almost perfect perception, i.e.: the few managers

reached by the word of mouth given by others,

perceive the advantage deriving from the new

technology. Thanks to the high perception, we have

the following increase rates: 11%, and then 12-14%

circa per year, for the following periods. After 5

simulated years, a total of 169.129 enterprises have

moved to the new technology, even if the network is

loosely connected, meaning that a lot of agents are

never reached by the news about technology. The

trend is again an increasing one, meaning that when

more agents are reached, there is in turn an higher

probability to reach other ones, and so on. The last

experiment is carried on with high perception and

highly connected network; now the spreading is

pervasive; after one year, we have an unrealistic

60% increase compared to the time zero, and the

trend continues in the following simulated years, so

that, after 60 months, 418.274 out of 500.000

enterprises are reached by the new technology.

5 CONCLUSIONS

The role of perception is investigated by studying

innovation diffusion. In particular, IT outsourcing is

a crucial innovation for enterprises and is used to

empirically vali-date the presented framework. An

agent based model is implemented, since it allows to

explore the human factor behind social phenomena.

Different experiments were carried on, by varying

the level of perception and of connection along the

social net-work. Validation for the results is carried

on by using tools coming from data-mining

(Remondino and Correndo, 2005). The results

obtained are straightforward: an high perception of

the advantages given by an innovation is crucial in

the spreading mechanism, and so is a good

connection level among the agents. By comparing

the real figures (coming from AITech-

Assinform/NetConsulting) about Italian market, we

see that they look like the ones obtained in the 1st

experiment. In fact, the delta in IT outsourcing

adoption from 2003 to 2004 is about 1%, stepping to

1.6% for 2004 to 2005, and to over 2% from 2005 to

2006. We conclude that the Italian SME market is

one in which there is a low perception of the benefits

and a loosely connected net-work.

ACKNOWLEDGEMENTS

We wish to acknowledge and thank prof. Anna

Maria Bruno (Dipartimento di Economia Aziendale,

University of Turin) for her precious and ongoing

support, as long as prof. Gianpiero Bussolin, our

unforgivable mentor.

REFERENCES

Remondino M., Correndo G., 2006. MABS Validation

Through Repeated Execution and Data Mining

Analisys. International Journal of Simulation. vol. 7,

pp. 10-21 ISSN: 1473-8031.

Rogers, E.M., 1983 Circulation of innovations. Int. J. Free

Press, New York

ICEIS 2009 - International Conference on Enterprise Information Systems

158