IAMSR & INTELLIGENT TECHNOLOGIES Christer Carlsson, Barbro Back and Pirkko Walden IAMSR / Åbo Akademi

IAMSR/Åbo Akademi University Academic research institute, funded by the Finnish industry [mainly through Tekes, EU-IST programs] Academic research institute, funded by the Finnish industry [mainly through Tekes, EU-IST programs] Network: City University of Hong Kong, TU Delft. Nankiang Technological Univ, UD (Dallas), UPMC (Paris), Univ of the Aegean, Univ of Trento, Univ of Vienna, … Network: City University of Hong Kong, TU Delft. Nankiang Technological Univ, UD (Dallas), UPMC (Paris), Univ of the Aegean, Univ of Trento, Univ of Vienna, … Corporate partners include Agentum, Amerpap, BTExact, Metso, MetsäTissue, M-real, Nokia, Finnforest, Outokumpu, Rautaruukki, Restel, TeliaSonera, UPM, Veritas, … Corporate partners include Agentum, Amerpap, BTExact, Metso, MetsäTissue, M-real, Nokia, Finnforest, Outokumpu, Rautaruukki, Restel, TeliaSonera, UPM, Veritas, … EUNITE Network of Excellence [EU-IST, smart adaptive systems] EUNITE Network of Excellence [EU-IST, smart adaptive systems] Berkeley Initiative in Soft Computing Berkeley Initiative in Soft Computing

IAMSR 2004 Working principles: Working principles: – we build theory, conceptual frameworks – we carry out interactive research processes – we do fundamental research – we carry out large projects and do project management – we do feasibility studies – we integrate systems from standard components

IAMSR/Åbo Akademi University Research staff 60+ of which 28 PhD students (8 countries) Research staff 60+ of which 28 PhD students (8 countries) – fuzzy logic, fuzzy optimization, soft computing – hyperknowledge and mobile support systems – intelligent software agents and approximate reasoning – e-commerce and m-commerce – neural nets and self-organizing maps, data mining – real options and fuzzy real options modeling – industry foresight with scanning, scenario agents – strategic management and scenario planning

IAMSR Structure IAMSR ASR KSRMSR Data Mining Mobile Commerce

IAMSR PROJECTS 2004 STEERING MCOMMERCE MCOMMERCE OptionsPorts , OptionsPorts , Phoenix Phoenix SoftLogs SoftLogs Rhodonea, 4M Rhodonea, 4M SmartBulls SmartBulls CHIMER CHIMER CoFI Projects CoFI Projects

GIGA-INVESTMENTS  Facts and observations  Giga-investments made in the paper- and pulp industry, in the heavy metal industry and in other base industries, today face scenarios of slow growth (2-3 % p.a.) in their key markets and a growing over-capacity in Europe  The energy sector faces growing competition with lower prices and cyclic variations of demand  Productivity improvements in these industries have slowed down to 1-2 % p.a

GIGA-INVESTMENTS  Facts and observations  Global financial markets make sure that capital cannot be used non-productively, as its owners are offered other opportunities and the capital will move (often quite fast) to capture these opportunities.  The capital markets have learned “the American way”, i.e. there is a shareholder dominance among the actors, which has brought (often quite short-term) shareholder return to the forefront as a key indicator of success, profitability and productivity.

GIGA-INVESTMENTS  Facts and observations  There are lessons learned from the Japanese industry, which point to the importance of immaterial investments. These lessons show that investments in buildings, production technology and supporting technology will be enhanced with immaterial investments, and that these are even more important for re-investments and for gradually growing maintenance investments.

GIGA-INVESTMENTS  Facts and observations  The core products and services produced by giga- investments are enhanced with lifetime service, with gradually more advanced maintenance and financial add-on services.  New technology and enhanced technological innovations will change the life cycle of a giga-investment  Technology providers are involved throughout the life cycle of a giga-investment

GIGA-INVESTMENTS  Facts and observations  Giga-investments are large enough to have an impact on the market for which they are positioned:  A ton paper mill will change the relative competitive positions; smaller units are no longer cost effective  A new teechnology will redefine the CSF:s for the market  Customer needs are adjusting to the new possibilities of the giga- investment  The proposition that we can describe future cash flows as stochastic processes is no longer valid; neither can the impact be expected to be covered through the stock market

GIGA-INVESTMENTS  The WAENO Lessons: Fuzzy ROV  Geometric Brownian motion does not apply  Future uncertainty [15-25 years] cannot be estimated from historical time series  Probability theory replaced by possibility theory  Requires the use of fuzzy numbers in the Black-Scholes formula; needed some mathematics  The dynamic decision trees work also with fuzzy numbers and the fuzzy ROV approach  All models could be done in Excel

REAL OPTIONS  Types of options  Option to Defer  Time-to-Build Option  Option to Expand  Growth Options  Option to Contract  Option to Shut Down/Produce  Option to Abandon  Option to Alter Input/Output Mix

REAL OPTIONS  Table of Equivalences: INVESTMENT OPPORTUNITY VARIABLE CALL OPTION Present value of a project’s operating cash flows S Stock price Investment costs X Exercise price Length of time the decision may be deferred t Time to expiry Time value of money rfrfrfrf Risk-free interest rate Risk of the project σ Standard deviation of returns on stock

REAL OPTION VALUATION (ROV) The value of a real option is computed by ROV =Se −δT N (d 1 ) − Xe −rT N (d 2 ) where d 1 = [ln (S 0 /X )+(r −δ +σ 2 /2)T] / σ√T d 2 =d 1 − σ√T,

FUZZY REAL OPTION VALUATION Fuzzy numbers (fuzzy sets) are a way to express the cash flow estimates in a more realistic way Fuzzy numbers (fuzzy sets) are a way to express the cash flow estimates in a more realistic way This means that a solution to both problems (accuracy and flexibility) is a real option model using fuzzy sets This means that a solution to both problems (accuracy and flexibility) is a real option model using fuzzy sets

FUZZY CASH FLOW ESTIMATES Usually, the present value of expected cash flows cannot be characterized with a single number. We can, however, estimate the present value of expected cash flows by using a trapezoidal possibility distribution of the form Usually, the present value of expected cash flows cannot be characterized with a single number. We can, however, estimate the present value of expected cash flows by using a trapezoidal possibility distribution of the form Ŝ 0 =(s 1, s 2, α, β)  In the same way we model the costs In the same way we model the costs

FUZZY REAL OPTION VALUATION We suggest the use of the following formula for computing fuzzy real option values Ĉ 0 = Ŝe −δT N (d 1 ) − Xe −rT N (d 2 ) where d 1 = [ln (E(Ŝ 0 )/ E(X))+(r −δ +σ 2 /2)T] / σ√T d 2 = d 1 − σ√T,

FUZZY REAL OPTION VALUATION E(Ŝ 0 ) denotes the possibilistic mean value of the present value of expected cash flows E(X) stands for the possibilistic mean value of expected costs σ: = σ(Ŝ 0 ) is the possibilistic variance of the present value of expected cash flows.

FUZZY REAL OPTION VALUATION  No need for precise forecasts, cash flows are fuzzy and converted to triangular or trapezoidal fuzzy numbers  The Fuzzy Real Option Value contains the value of flexibility

FUZZY REAL OPTION VALUATION

SCREENSHOTS FROM MODELS

NUMERICAL AND GRAPHICAL SENSITIVITY ANALYSES

FUZZY OPTIMAL TIME OF INVESTMENT Ĉ t* = max Ĉ t = Ŵ t e -δt N(d 1 ) – X e -rt N (d 2 ) t =0, 1,...,T where Ŵ t = PV(ĉf 0,..., ĉf T, β P ) - PV(ĉf 0,..., ĉf t, β P ) = PV(ĉf t +1,..., ĉf T, β P ) Invest when FROV is at maximum:

OPTIMAL TIME OF INVESTMENT C t* = max C t = V t e -δt N(d 1 ) – X e -rt N (d 2 ) t =0, 1,...,T How long should we postpone an investment? Benaroch and Kauffman (2000) suggest: Optimal investment time = when the option value C t* is at maximum (ROV = C t* ) Where V t = PV(cf 0,..., cf T, β P ) - PV(cf 0,..., cf t, β P ) = PV(cf t +1,...,cf T, β P ),

FUZZY OPTIMAL TIME OF INVESTMENT We must find the maximising element from the set {Ĉ 0, Ĉ 1, …, Ĉ T }, this means that we need to rank the trapezoidal fuzzy numbers In our computerized implementation we have employed the following value function to order fuzzy real option values, Ĉ t = (c t L,c t R,α t, β t ), of the trapezoidal form: v (Ĉ t ) = (c t L + c t R ) / 2 + r A · (β t + α t ) / 6 where r A > 0 denotes the degree of the investor’s risk aversion

1  A B AxBAxB Non-interactive possibility distributions

1  A B C The case of  f (A, B) = 1

1  A B D The case of  f (A, B) = –1

1  A B E The case of  f (A, B) = 1/3

1  A B F The case of  f (A, B) = –1/3

1  A B I The case of interactivity when  f (A, B) = 0

1  B AC